Radiotheranostic Agent ⁶⁴Cu-cyclam-RAFT-c(-RGDfK-)₄ for Management of Peritoneal Metastasis in Ovarian Cancer

Epithelial ovarian cancer accounts for more than 90% of ovarian malignancies, typically occurs in postmenopausal women, and remains the leading cause of death from gynecologic malignancies (1). Ovarian cancer is a highly metastatic disease characterized by widely disseminated tumor depositions throughout the peritoneal cavity together with malignant ascites (2). Because of the lack of clear symptoms or an effective screening method in the early stage of the disease, more than 70% of patients with ovarian cancer are diagnosed with peritoneal metastases (PM) and have a 5-year survival rate of <30% (1, 2). Standard care for ovarian cancer peritoneal metastasis (OCPM) consists of cytoreductive surgery followed by courses of chemotherapy. Unfortunately, most patients ultimately develop recurrent chemoresistance after multiple relapses, resulting in a dismal prognosis (3). Although ovarian cancer cells are radiosensitive (4), conventional external beam radiotherapy is typically not suited for OCPM because of the high toxicity of wide-field (whole-abdomen) irradiation (5). Therefore, there is a great demand for the development of new targeted therapeutic options for OCPM (6).

α_Vβ₃ integrin (α_Vβ₃) is a transmembrane cell adhesion receptor and a receptor for the well-known cyclic Arg-Gly-Asp sequence–containing pentapeptide (cRGD) ligands (7). It is overexpressed on tumoral neoendothelial cells during angiogenesis and frequently on cancerous cells, where it regulates angiogenesis and has roles in tumor proliferation, invasion, and metastasis (8, 9). Elevated expression of α_Vβ₃ has been reported in ovarian cancer cell lines (10, 11) and in approximately 80% of human ovarian cancer specimens (12, 13). However, the use of α_Vβ₃ inhibitors (antibody and peptide) to block the α_Vβ₃-associated pathways has had limited effect for ovarian cancer treatment (14). This has also been the case with other molecular-targeted therapy studies, perhaps because multiple key pathways are involved in cancer progression and can compensate when a single pathway is blocked (14, 15).

Targeted radionuclide therapy (TRT) is an evolving cancer treatment modality based on systemic administration of a specific carrier (antibody, peptide, or small molecule) conjugated with a cytocidal energy-releasing radionuclide (emitting α or β⁻ particles or Auger electrons) for carrier-guided irradiation of both primary and metastatic lesions (16). Peptide-based carriers may offer key advantages, such as rapid blood clearance, high tissue penetration, and low immunogenicity (17). Several groups, including our own, have demonstrated the therapeutic potential of TRT based on the cRGD ligand and α_Vβ₃ receptor (α_Vβ₃-TRT) using subcutaneous mouse models of several cancer types (18–20), but little work has been done using the OCPM model. In 2007, Dijkgraaf and colleagues reported a survival-prolonging effect for the ¹⁷⁷Lu-labeled monomeric cRGD peptide in the mouse NIH:OVCAR-3 (OVCAR-3) OCPM model (21). They started the treatment on ascites cell clusters and small tumor depositions (but provided no information on tumor size). An ¹¹¹In-labeled counterpart was found to accumulate mainly in the periphery of the examined tumors, demonstrating the heterogeneous intratumoral distribution (ITD) of this cRGD monomer.

⁶⁴Cu-cyclam-RAFT-c(-RGDfK-)₄ (⁶⁴Cu-RaftRGD; Fig. 1A) is a tetrameric cRGD peptide with high α_Vβ₃ binding affinity and specificity that is armed with a promising theranostic radionuclide. ⁶⁴Cu has a suitable half-life (12.7 hours) and multiple decay modes, including β⁺ emission, which can be used for PET, and the combined emission of β⁻ and Auger electrons, which can be used for therapeutic irradiation (22–24). We have previously demonstrated the use of ⁶⁴Cu-RaftRGD PET to successfully visualize and quantify α_Vβ₃ expression in mice bearing subcutaneous xenografts of U87MG glioblastoma and HuH-7 hepatocellular carcinoma, as well as mice bearing an orthotopic xenograft of BxPC-3 pancreatic cancer (24–27). We subsequently showed the therapeutic potential of ⁶⁴Cu-RaftRGD and its ⁶⁷Cu (β⁻) analog in the U87MG model (20, 28). Here, we evaluated the utility of ⁶⁴Cu-RaftRGD in a clinically relevant α_Vβ₃-positive OCPM mouse model by determining the optimal route of drug administration and investigating both its PET compatibility and therapeutic potential (in terms of radiation dosimetry, antitumor activity, and effects on normal tissues). Given the heterogeneous nature of OCPM lesions, we also studied the ITD and intertumoral distribution of ⁶⁴Cu-RaftRGD in relation to selected components of the tumor microenvironment.

Cyclam-RAFT-c(-RGDfK-)₄ [molecular weight (MW): 4,119.6; Fig. 1A], Cy5.5-conjugated RAFT-c(-RGDfK-)₄ (Cy5.5-RaftRGD; MW: 5,234.3; Cy5.5: emission maximum, 694 nm; Fig. 1A), and RAFT-c(-RGDfK-)₄ (RaftRGD; MW: 3,879.3) were synthesized as described previously (22, 23, 29). ⁶⁴CuCl₂ (radionuclide purity, >99.9%) and ⁶⁴Cu-RaftRGD (maximal molar activity, 740 MBq/nmol and radiochemical purity, >99%) were prepared on the basis of methods reported previously (24, 30, 31).

Two human epithelial ovarian adenocarcinoma cell lines were used. IGR-OV1 (Lot number 0509653) was obtained through a material transfer agreement (#1-5734-18) from the National Cancer Institute, Division of Cancer Treatment and Diagnosis (DCTD) Tumor Repository, NCI-Frederick, Frederick National Laboratory for Cancer Research (Frederick, MD; 25 May 2018; authenticated by Applied Biosystems AmpFLSTR Identifiler testing). OVCAR-3 was directly obtained from the ATCC (on 15 March 2018 and characterized by DNA profile and cytogenetic analysis). Cells were cultured according to the manufacturer's instructions. It should be stated that upon arrival (passage 0), the two lines were expanded and frozen in our laboratory (no further authentication and Mycoplasma testing were performed), and early passage cells (passages 6–7 and 11–14 for IGR-OV1 and OVCAR-3, respectively) were used for all experiments.

All animal studies were approved by the Animal Ethics Committee of the National Institutes for Quantum and Radiological Science and Technology (Chiba, Japan) and were conducted in accordance with institutional guidelines. Four- and 7-week-old female BALB/cAJcl-nu/nu mice were obtained from CLEA Japan for tumor-bearing and normal (nontumor bearing) mice studies, respectively, and acclimated for 1 week before experimentation. The mice were injected intraperitoneally with the designated number of IGR-OV1 or OVCAR-3 cells, and tumor development was determined by necropsy together with examination of ascites using a Leica DMI 4000B Inverted Microscope (Leica Microsystems). For subcutaneous tumor reference, mice were inoculated with 5 × 10⁶ IGR-OV1 or 1 × 10⁷ OVCAR-3 cells in the right flank and used for experiments when tumors were 6–9 mm in length (2–6 weeks after xenografting).

Flow cytometry and IHC for α_Vβ₃ expression were performed using methods reported previously (10, 20), which are briefly described in the Supplementary Data.

Cy5.5-RaftRGD (10 nmol in 0.2-mL PBS) was administered intravenously (via the tail vein) or intraperitoneally to OCPM-bearing mice and intravenously to mice bearing IGR-OV1 subcutaneous tumors. Mice were euthanized at 2 hours postinjection and selected tumors were removed and cryosectioned (10-μm thick) as described previously (26). Air-dried sections were simultaneously scanned (at 700 nm and 21-μm resolution) using an Odyssey CLx NIR Imaging System (LI-COR Biosciences) to visualize Cy5.5-RaftRGD distribution as described previously (31). Afterwards, sections were subjected to CD31 staining as described previously (26) with anti-CD31 (1:1,500 dilution; BD Pharmingen) and IRDye 680CW–conjugated goat anti-rat IgG (1:200 dilution; LI-COR Biosciences) to visualize vasculature. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Stained sections were scanned with the Odyssey as mentioned above. DAPI staining was imaged using a Keyence BZ-X710 Fluorescence Microscope (Keyence). It should be noted that the fluorescence intensity of Cy5.5-RaftRGD was remarkably weakened during the immunostaining process, and thus, did not interfere with the detection of CD31 staining.

OCPM-bearing mice received sequential intravenous or intraperitoneal coadministration of ⁶⁴Cu-RaftRGD [37 MBq/0.1 nmol in 0.2-mL normal saline with 1% Tween 80 (NS/Tw80), the vehicle for ⁶⁴Cu-RaftRGD; ref. 20] followed by 10 nmol Cy5.5-RaftRGD at an interval of 10–20 minutes to avoid competitive receptor blocking. Three hours after the first injection, mice were euthanized. Cryosections of selected tumors with adjacent small intestine were sequentially subjected to autoradiography as described previously (refs. 26, 31; briefly described in the Supplementary Materials and Methods), Odyssey scanning as described above, and hematoxylin and eosin (H&E) staining. Adjacent sections were subjected to CD31/DAPI staining as described above and imaged using the Keyence microscopy. Tumors of untreated mice were subjected to Odyssey scanning and H&E staining to discriminate background fluorescence (autofluorescence). ⁶⁴CuCl₂ (37 MBq in 0.2-mL PBS) was injected either intravenously or intraperitoneally into another set of OCPM-bearing mice, and selected tumors at 3 hours postinjection were subjected to autoradiography and CD31/DAPI/H&E staining.

OCPM-bearing mice received concurrent intraperitoneal coadministration of 10 nmol Cy5.5-RaftRGD and 1.5-mg pimonidazole (Pimo) hydrochloride (Hypoxyprobe-1 Omni Kit; Hypoxyprobe) and euthanized at 2 hours postinjection. Tumor cryosections were sequentially subjected to: (i) fixation with 2% paraformaldehyde and DAPI staining, (ii) microscopy to visualize Cy5.5-RaftRGD distribution, (iii) double staining by coincubation with anti-CD31 (1:200 dilution) and rabbit anti-Pimo (1:100 dilution; Hypoxyprobe-1 omni kit) followed by Alexa Fluor 594–conjugated goat anti-rat IgG (1:100 dilution; Invitrogen) and IRDye 800CW–conjugated goat anti-rabbit IgG (1:100 dilution; LI-COR Biosciences), and (iv) microscopy to visualize CD31 and Pimo staining. Microscopic images of each section in its entirety were captured using a 20× objective on the Keyence microscope with DAPI, TRITC (for CD31 staining), Cy5.5, and Cy7 (for Pimo staining) filters using identical settings for resolution and capture position. On adjacent sections, α_Vβ₃-IHC was additionally performed for reference.

Similar methods were used in mice bearing IGR-OV1 or OVCAR-3 subcutaneous tumors receiving intravenous injection of Cy5.5-RaftRGD and Pimo. See the Supplementary Data for the detailed procedures.

PET, CT, and contrast-enhanced CT (CECT) for the designated groups of mice were carried out according to our methods reported previously (25, 27), which are briefly described in the Supplementary Data. All scans (mice in prone position) were displayed as decay-corrected maximum-intensity projection (MIP)-PET images or as fused PET/CT or CECT images using Inveon Research Workplace Software (version 4.0, Siemens). Quantification was performed on coronal PET images, which is briefly described in the Supplementary Data. All mice were euthanized immediately after the scans (∼24 hours postinjection), after which selected tumors and major organs were harvested for activity measurement with decay correction on a Gamma Counter (2480 WIZARD², PerkinElmer). For ascites, cells were spun at 9,100 × g for 2 minutes, and radioactivity was measured in the cell pellet and supernatant. Data are expressed as the percentage injected radioactivity dose (%ID) per gram tissue or per milliliter ascites (%ID/g or %ID/mL; normalized to a mouse of 20 g body weight).

In the first experiment, normal mice (n = 3–5/group) were injected intraperitoneally with 0.74 MBq/0.2 nmol ⁶⁴Cu-RaftRGD and euthanized at 2 minutes, 20 minutes, 1 hour, 3 hours, 24 hours, and 48 hours postinjection. Organs of interest were harvested for %ID/g quantification. Absorbed doses (ADs) were estimated using the residence time represented by the AUC of organ activity versus time and the mean energy emitted per transition of ⁶⁴Cu (1.22 × 10⁻¹⁴ Gy kg/Bq/s) as described previously (20).

In the second experiment, a similar assay was performed in OCPM-bearing mice (n = 3/group). For a blocking experiment, mice (n = 3) were coinjected with a 100-fold excess of unlabeled RaftRGD (20 nmol) and examined at 3 hours postinjection. Blood, ascites, and tumors of varying size (weight and length) were assessed.

OCPM-bearing mice (n = 2 or 3/group) were injected intraperitoneally with NS/Tw80 (vehicle control) or ⁶⁴Cu-RaftRGD (148 MBq/0.357 nmol) and euthanized after 3 days. Tumors of varying size, small intestine, and kidney were dissected and cryosectioned for H&E staining and immunostaining for proliferation (Ki-67) and apoptosis [terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) assay]. The procedures, based on our methods reported previously (20), are described in detail in the Supplementary Data.

Survival study (n = 9/group) and toxicity evaluation (n = 5/group) were additionally performed in OCPM-bearing and normal mice, respectively. See details in the Supplementary Data.

Quantitative data are presented as the mean ± SD. Student t tests (unpaired and two-tailed) were used for two-group comparisons, and a one-way ANOVA with Dunnett test was used for multiple group comparisons (KaleidaGraph 4.0). Correlations were determined using Pearson correlation test (GraphPad Prism 7.04). P < 0.05 was considered significant.

Figure 1B demonstrates expression of α_Vβ₃ in approximately 10% of OVCAR-3 cells and >50% of IGR-OV1 cells. Both lines induced tumor formation when inoculated intraperitoneally or subcutaneously into athymic nude mice, with IGR-OV1 cells displaying higher tumorigenicity. On days 52–67 after intraperitoneal injection of 4 × 10⁷ OVCAR-3 cells, no visible abdominal distention was observed, and three of the four mice (75%) formed primarily milky and viscous ascites that contained large amounts of cancer cell clusters (Fig. 1C; Supplementary Fig. S1A) and small numbers (five or fewer) of tumor deposits in the peritoneal cavity (Fig. 1C). In contrast, following intraperitoneal injection of 1 × 10⁷ IGR-OV1 cells, four of four mice (100%) developed visibly distended abdomens, copious yellowish or bloody ascites containing a high density of single cancer cells and cancer cell clusters (Fig. 1D; Supplementary Fig. S1B), and dozens of various sized tumor lesions disseminated throughout the peritoneal cavity (Fig. 1D) within 14–18 days. IHC staining with anti-human α_Vβ₃ showed strong membranous staining of tumor cells in mice receiving OVCAR-3 and IGR-OV1 intraperitoneal injections (Fig. 1C and D) or subcutaneous xenografts (Supplementary Fig. S2). Intratumoral heterogeneous expression of α_Vβ₃ was also obvious, with the greatest staining in perivascular tumor cells or the periphery of tumor cell nests, where tumor cells were in contact with surrounding tissues. On the basis of these findings, we used mice inoculated intraperitoneally with IGR-OV1 cells (∼18 days after xenografting) as the main OCPM model (if not specified otherwise) in this study. These mice had the typical OCPM presentation described above, with no visible signs of compromised health.

To predict the ITD of ⁶⁴Cu-RaftRGD, we analyzed its fluorescence surrogate, Cy5.5-RaftRGD, at 2 hours after intravenous and intraperitoneal injection into mice bearing IGR-OV1 OCPM (IGR-OV1–OCPM mice). Figure 2 shows representative images of the distribution of Cy5.5-RaftRGD (Fig. 2A) and corresponding blood vessels (Fig. 2B) in entire sections of OCPM lesions (a cluster of small nodules and a single large nodule) after intravenous and intraperitoneal injection, and reference images from an IGR-OV1 subcutaneous tumor after intravenous injection. In all tumors examined, there was no demonstrable difference in Cy5.5-RaftRGD uptake between the tumor center and periphery, which thus, resulted in a whole-tumor binding pattern. Under comparable conditions (in terms of tumor size and vasculature), peritoneal tumors labeled with an intraperitoneal injection showed a generally higher overall uptake than those labeled with an intravenous injection. Comparison of large nodules demonstrated that the only notable difference in Cy5.5-RaftRGD distribution between intravenous and intraperitoneal injection routes was a higher uptake at the tumor periphery following intraperitoneal injection. This is mostly attributable to direct penetration of the probe from the surrounding ascites. Comparison of the intravenous injection between peritoneal and subcutaneous tumors indicated that the lower uptake related to peritoneal tumors might be at least, in part, due to the compromised vasculature that develops in the peritoneal cavity as compared with the highly vascularized subcutaneous xenograft (Fig. 2B).

Intratumoral colocalization of Cy5.5-RaftRGD and ⁶⁴Cu-RaftRGD was validated by Odyssey scanning and autoradiography of the same whole-tumor sections at approximately 3 hours after intravenous or intraperitoneal coadministration to IGR-OV1–OCPM mice. Figure 3 shows representative images of the well-matched and overlapping distribution of Cy5.5-RaftRGD and ⁶⁴Cu-RaftRGD in clusters of small tumors after intravenous or intraperitoneal injection. In contrast to intravenous injection, intraperitoneal injection led to uptake of ⁶⁴Cu/Cy5.5-RaftRGD in avascular or poorly vascularized tumor lesions. The ITD of free ⁶⁴CuCl₂ (Supplementary Fig. S3), injected either intravenously or intraperitoneally, demonstrated poor tumor selectivity (compared with adjacent intestinal tissues) in the current OCPM model. Moreover, intraperitoneally administered ⁶⁴CuCl₂ resulted in accumulation of radioactivity that was predominantly in the peripheral rim of the tumor, forming a ring-like appearance.

Next, the ITD of Cy5.5-RaftRGD (representing ⁶⁴Cu-RaftRGD) was analyzed in relation to CD31-stained blood vessels and regions of Pimo-stained hypoxic cells by high-resolution multifluorescence microscopy in tumor sections at 2 hours after coadministration of the probes to IGR-OV1–OCPM mice. Figure 4A–J shows a representative OCPM section exhibiting separate and merged images of nuclei, Cy5.5-RaftRGD, microvessels, and regions of hypoxia. In Fig. 4D–F, regions of partial overlap (yellow) between Cy5.5-RaftRGD (red; Fig. 4B) and microvessels (green; Fig. 4C) denote α_Vβ₃-overexpressing neovessels. In Fig. 4H–J and Supplementary Fig. S4B, merged images of Cy5.5-RaftRGD and hypoxia (green; Fig. 4G) clearly demonstrate the discordant, but complementary, distributions of Cy5.5-RaftRGD and hypoxia. The overlapping signal at the tumor periphery is most likely attributable to direct penetration of the two intraperitoneally injected probes from the surrounding ascites. Similar results were obtained in IGR-OV1 and OVCAR-3 subcutaneous tumor models (Fig. 4K–N and more details in Supplementary Fig. S5).

Figure 5 shows representative whole-body MIP-PET and magnified abdominal MIP-PET/CT (or CECT) images after intravenous or intraperitoneal injection of 0.2 nmol ⁶⁴Cu-RaftRGD (∼18.5 or 37 MBq) in tumor-bearing and normal mice. Here, and below, this 0.2-nmol dose was supplemented by the addition of unlabeled cyclam-RAFT-c(-RGDfK-)₄ to be the equivalent of a 148-MBq therapy plan. The intravenous injection in IGR-OV1–OCPM mice resulted in rapid blood clearance and notable renal excretion as early as 5 minutes after injection, with generally low accumulation of abdominal radioactivity and tumors that were barely detectable (Fig. 5A, D, and E). In intraperitoneally treated IGR-OV1–OCPM mice (Fig. 5B, F, and G) and normal mice (Fig. 5C, H, and I), high levels of radioactivity accumulated throughout the abdominal area within the first 1 hour postinjection. After 3.5 and 24 hours postinjection, this noticeable background declined significantly, allowing delineation of accumulation within the renal excretion system. Meanwhile, multiple focal regions of high radioactivity uptake were detected only in the abdominal cavity of tumor-bearing mice, indicating the sites of metastases. These findings were validated by the results of quantitative PET (Supplementary Fig. S6A), post-PET biodistribution (Supplementary Fig. S6B; Supplementary Table S1), and time-course biodistribution given in the following section.

Additional PET/CT or PET/CECT of intraperitoneally treated mice bearing OVCAR-3 OCPM (OVCAR-3–OCPM mice) showed strong radioactivity signals in regions of ascites high accumulation (associated with the body position; Fig. 5J–L), with high values of 11.2 ± 1.99 and 8.3 ± 2.05 %ID/g_max (maximal %ID/g) at 3.5 and 24 hours postinjection, respectively (Supplementary Fig. S7A). Post-PET biodistribution (∼24 hours postinjection) revealed 98.8% ± 0.15% of ascites radioactivity attributed from cancer cell clusters (as seen in Supplementary Fig. S1A) with a high uptake of 16.2 ± 2.92 %ID/g (Supplementary Fig. S7B and S7C).

The biodistribution results of intraperitoneally administered ⁶⁴Cu-RaftRGD (⁶⁴Cu-RaftRGD; 0.74 MBq/0.2 nmol, i.p.) in normal mice are summarized in Supplementary Fig. S8 and Supplementary Table S2. In the blood, radioactivity peaked (0.59 ± 0.09 %ID/g) at 1 hour postinjection and dropped to a negligible level (0.04 ± 0.01 %ID/g) at 24 hours postinjection. The kidneys had the highest uptake (24.8 ± 2.14 and 18.2 ± 1.59 %ID/g at 3 and 24 hours postinjection, respectively), indicating prominent renal excretion. The estimated mean ADs in blood and major abdominal organs are also presented in Supplementary Fig. S8 (human dosimetry is appended in Supplementary Table S4 for reference). The kidneys received the highest dose (24.6 Gy when injected with 148 MBq/0.2 nmol ⁶⁴Cu-RaftRGD). Doses to nonabdominal organs (such as heart and lungs) can be inferred from our previous data for intravenously administered ⁶⁴Cu-RaftRGD (20), as the PET (Fig. 5A–C) and post-PET biodistribution (Supplementary Fig. S6B) analyses demonstrated similarly low accumulation levels of radioactivity in these organs after intravenous and intraperitoneal injection.

The results of blood and ascites obtained from IGR-OV1–OCPM mice are summarized in Supplementary Figs. S9 and S10A–S10G, respectively, and those of tumors are shown in Fig. 6A and B and Supplementary Figs. S11 and S12. There was little to no difference in blood radioactivity levels and kinetics between tumor-bearing and normal mice (Supplementary Fig. S9). The concentration of radioactivity in ascites (Supplementary Fig. S10A) declined steadily over time, from 40.6 ± 10.7 and 31.4 ± 3.6 %ID/mL at 2 and 20 minutes postinjection, respectively, to 17.4 ± 1.9, 6.3 ± 0.2, and 0.89 ± 0.12 %ID/mL at 1, 3, and 24 hours postinjection, respectively. This result is consistent with the changes in abdominal radioactivity as visualized by PET (Fig. 5B). Uptake in the cell pellet fraction from ascites peaked at 3 hours postinjection (13.0 ± 2.2 %ID/g; Supplementary Fig. S10F) and decreased gradually to 5.1 ± 0.47 %ID/g at 48 hours postinjection. Because the ascites cell pellets of IGR-OV1–OCPM mice also contained large numbers of blood cells (Supplementary Fig. S1B), these data can be considered underestimate of the level of uptake by the ascites cancer cells. Uptake of radioactivity was then evaluated in 162 tumors (nine tumors of varying size/mouse; Supplementary Fig. S11) that had a r value of 0.920 between tumor weight and length (P < 0.0001; Supplementary Fig. S12). Similar to the cancer cell clusters in ascites, tumor deposits had a peak uptake at 3 hours postinjection. The mean peak uptake of the smallest tumors (≤2.6 mm; 10 ± 1.9 %ID/g) was nearly two times that of the largest tumors (≥10 mm; 5.3 ± 1.0 %ID/g, P = 0.0003). Figure 6A shows scatter plots and fitted curves of tumor uptake versus length at multiple timepoints, revealing a significant inverse correlation for each timepoint. Size-specific tumor uptake values were predicted at the highest values of r using the inverse regression relationship expressed by the equation |y = {\frac{a}{x}} + b$| (Supplementary Table S3). Next, size-dependent estimated ADs were obtained (Fig. 6B). Finally, coinjection with an excess of unlabeled RaftRGD resulted in a marked decrease in the uptake of ⁶⁴Cu-RaftRGD by both tumor nodules (irrespective of size; Fig. 6A; Supplementary Fig. S11) and ascites cell pellets (Supplementary Fig. S10G), revealing an α_Vβ₃-specific binding mechanism.

Figure 6C and D show the effects of intraperitoneal injection of ⁶⁴Cu-RaftRGD (148 MBq/0.357 nmol) on proliferation and apoptosis in tumors of IGR-OV1–OCPM mice, respectively. The peptide dose of 0.357 nmol was slightly higher than the planned dose (0.2 nmol) because of the molar activity of 415 MBq/nmol obtained from the radiolabeling. It should be noted that, as RaftRGD itself has no antitumor activity even at a dose of 4 nmol (28), a peptide control was not used in this therapy study. Tumors from both vehicle- and ⁶⁴Cu-RaftRGD–treated mice ranged from a few hundreds of micrometers to >10 mm in length and demonstrated highly varied Ki-67 proliferation indices (percentage of cells positive for Ki-67) and apoptotic indices (percentage of apoptotic cells). To minimize the effect of tumor size on changes in proliferation and apoptosis, tumors were categorized into three groups on the basis of tumor length and number of tumors (Fig. 6C and D): <1 mm, 1–4 mm, and >4 mm. ⁶⁴Cu-RaftRGD reduced proliferation by 47.9%, 62.8%, 40.9%, and 27.9% and increased apoptosis by 3.2-, 4.8-, 3.0-, and 2.7-fold in comparisons among all lesions and lesions <1 mm, 1–4 mm, and >4 mm, respectively. Supplementary Figure S13 shows the results of proliferation/morphology studies on small intestine and kidney. One major finding was that the proliferation index in kidney was lower in ⁶⁴Cu-RaftRGD–treated mice than in control mice (0.26% ± 0.07% vs. 1.72% ± 0.15%; P = 0.0068).

The long-term antitumor activity of the single-dose of 148 MBq/0.357 nmol ⁶⁴Cu-RaftRGD i.p. was revealed by the survival study (summarized in Supplementary Fig. S14A–S14F). Ascites formation was markedly suppressed (Supplementary Fig. S14E), and in-line with the Ki-67 and apoptosis analyses described above, the smaller sized tumors (like those along the intestine) appeared to be more efficiently treated (Supplementary Fig. S14A and S14F).

The results of toxicity evaluation of 148 MBq/0.357 nmol ⁶⁴Cu-RaftRGD i.p. in normal mice as long as 60 days after administration are given in Supplementary Figs. S15A–S15J and S16A–S16C. No severe toxicity occurred, as the most noticeable findings included a mild transient decrease of blood counts and slightly elevated levels of plasma creatinine on day 60.

OCPM presents with a pathophysiologically heterogeneous group of tumors, typically with the coexistence of ascites cancer cell clusters and solid tumors of varying size disseminated throughout the peritoneal cavity. To be effective and informative, a new TRT strategy for OCPM should be developed on the basis of an in-depth understanding of the relevant intratumoral heterogeneity and intertumoral heterogeneity. This study shows the theranostic potential of an α_Vβ₃-targeted multimeric cRGD peptide–based radiopharmaceutical agent in clinically relevant mouse models of OCPM (IGR-OV1 and OVCAR-3), along with analysis of the intratumoral heterogeneity and intertumoral heterogeneity of both drug uptake and therapeutic effect. Importantly, it also demonstrates a spatially complementary relationship between the ITD of the α_Vβ₃-targeting agent and regions of tumor hypoxia.

Determining the optimal administration route (intravenous or intraperitoneal injection) to achieve efficient targeting to PMs has been a long-standing issue that needs to be addressed (21, 32). Because intraperitoneal injection can deliver drug to the tumor not only via the blood circulation (through absorption by mesenteric and lymphatic vessels), but also via direct penetration, it can result in increased drug concentrations in the tumor (33, 34). However, the outcome will also depend on the collective effects of a number of other factors, such as the efficiency of drug diffusion inside the peritoneal cavity upon injection, the capability of the drug to penetrate the tumor, the metabolic stability of the drug in ascites fluid, and the extent of drug binding to ascites or tissue proteins. This study demonstrates that the tumor-targeting efficiency of ⁶⁴Cu-RaftRGD in OCPM-bearing mice was markedly influenced by the route of administration. A direct comparison showed that the intraperitoneal route produced high levels of tumor uptake (Supplementary Fig. S6) in contrast to the less efficient intravenous route, enabling PET visualization of multiple regions of metastases (both of solid tumors and malignant ascites) in the peritoneal cavity (Fig. 5). Moreover, the observed whole-tumor binding patterns of intraperitoneally administered Cy5.5/⁶⁴Cu-RaftRGD observed within a few hours after injection (2–3 hours postinjection; Figs. 2–4) indicate that RaftRGD-based probes have an excellent ability to penetrate tumors.

α_Vβ₃-specific tumor uptake of intraperitoneally injected ⁶⁴Cu-RaftRGD was validated by coinjection of an excess of unlabeled parent RaftRGD peptide, which resulted in a significant decrease in accumulation of radioactivity in both solid tumor deposits (Fig. 6A; Supplementary Fig. S11) and ascites cell pellets (Supplementary Fig. S10E and S10G). Moreover, the ITD of free ⁶⁴CuCl₂ administered intraperitoneally or intravenously showed poor tumor selectivity, with a predominant ring-like distribution at the tumor periphery after intraperitoneal injection (Supplementary Fig. S3), indicating poor tumor penetration. The distinct differences between the binding patterns of ⁶⁴Cu-RaftRGD and ⁶⁴CuCl₂ demonstrate the stability of ⁶⁴Cu-RaftRGD in ascites fluid, with no major release of ⁶⁴Cu from the probe. Notably, the distribution of ⁶⁴CuCl₂ indicates that it was incapable, at least in this case, of specifically targeting PMs, although some recent preclinical studies have demonstrated its theranostic potential for several other cancer types, including melanoma, glioblastoma, and prostate cancer as reviewed by Gutfilen and colleagues (35).

We observed an inverse correlation between tumor uptake of intraperitoneally injected ⁶⁴Cu-RaftRGD and tumor size (1.2–17.2 mm) at all examined timepoints (2 minutes to 48 hours postinjection; Fig. 6A), consistent with other reports on different radiocompounds and tumor models (36–38). This indicates an intertumoral heterogeneity, in which small tumors received proportionally higher radiation doses. This size effect can be affected by a number of other factors, such as changing levels of target density or activity or the degree of tumor vascularity, hypoxia, and necrosis. In the current case, hypoxia appeared to be closely related to ⁶⁴Cu-RaftRGD tumor uptake, based on the spatially complementary ITD patterns of compound and hypoxia (described further below) and an increase in the severity of hypoxia with increasing tumor size (Supplementary Fig. S17A). On the basis of the correlation between intraperitoneally injected ⁶⁴Cu-RaftRGD tumor uptake and tumor size, we developed inverse regression equations for each timepoint postinjection (Supplementary Table S3), allowing us to predict tumor uptake based on tumor size and hence, calculate an estimated size-dependent AD (Fig. 6B).

Intertumoral heterogeneity was also found with regard to the antitumor activity of intraperitoneally injected ⁶⁴Cu-RaftRGD against OCPMs of various sizes. Antitumor activity was evaluated on the basis of the ability of intraperitoneally administered compound to inhibit tumor proliferation (Fig. 6C) and induce apoptosis (Fig. 6D), leading to a marked suppression of ascites formation (Supplementary Fig. S14E) and tumor growth (Supplementary Fig. S14F). A rough association was seen between tumor size and therapeutic response. At ADs of >28.2, 28.2–8.8, and <8.8 Gy to tumors <1 mm, 1–4 mm, and >4 mm (data derived from Fig. 6B), we observed a 62.8%, 40.9%, and 27.9% reduction in proliferation index and a 4.8-, 3.0-, and 2.7-fold increase in apoptotic index, respectively. Given the heterogenic effects of this therapy, our findings suggest approaches for optimizing treatment parameters based on the extent of metastasis (such as the tumor size).

Tumor heterogeneity, including that of drug distribution, is an inevitable obstacle limiting treatment efficacy (39, 40). In TRT, the heterogeneous ITD of radiocompounds, at least, in part, due to the intrinsically heterogeneous ITD of targets (e.g., different antigen expression levels among cancer cells), as well as the influence of the tumor microenvironment, can cause nonuniform delivery of cytocidal radiation throughout the tumor mass, resulting in a nonuniform therapeutic response. In this study, we investigated the ITD of ⁶⁴Cu-RaftRGD in relation to two important features of the solid tumor microenvironment: tumor vasculature and hypoxia (41). Using the surrogate probe, Cy5.5-RaftRGD, in our unique high-resolution multifluorescence analysis, we found a spatially discordant, but complementary, relationship between the ITD of Cy5.5-RaftRGD (in regions of angiogenesis and α_Vβ₃-positive cancerous cells) and regions of hypoxia (where α_Vβ₃ expression itself can be found positive as shown in Supplementary Fig. S4C) in the OCPM model as well as subcutaneous IGR-OV1 and OVCAR-3 models (Fig. 4). This striking finding may lay out a rationale for pairing α_Vβ₃- and hypoxia-targeted TRT agents to tackle the obstacle of heterogeneity encountered by single TRT agents. Indeed, we recently reported that the combined use of ⁶⁴Cu-RaftRGD and ⁶⁴Cu-diacetyl-bis (N⁴-methylthiosemicarbazone) (⁶⁴Cu-ATSM, a supposed tracer for hypoxia metabolism) in U87MG subcutaneous xenografts improves the antitumor effect as compared with ⁶⁴Cu-RaftRGD alone, owing to a more uniform ITD of radioactivity (31).

The kidneys are dose-limiting organs for radiotherapy. Here, in mice receiving 148 MBq ⁶⁴Cu-RaftRGD i.p., the kidneys had an estimated mean AD of 24.6 Gy. Recent peptide-based studies (42, 43) comparing the nephrotoxicity of ⁹⁰Y- and ¹⁷⁷Lu-radiolabeled somatostatin analogues propose the renal dose limits of 24 Gy for ⁹⁰Y and a higher value of 28 Gy for ¹⁷⁷Lu (due to a shorter range β⁻ particles, 2 mm maximum than ⁹⁰Y, ∼11.3 mm maximum, causing less damage to nearby tissue). Because ⁶⁴Cu has a similar β⁻ particle range (2.5 mm maximum; ref. 44) with ¹⁷⁷Lu, it is reasonable to postulate a renal dose limit of approximately 28 Gy for ⁶⁴Cu-RaftRGD. In fact, the 60-day assessment demonstrates no evidence of severe toxicity in the 148-MBq–treated mice (Supplementary Figs. S15 and S16). On the other hand, several approaches have been introduced to prevent renal damage from high ADs of radiation (44). We have previously shown that introducing gelofusine (succinylated gelatin) and l-lysine (positively charged basic amino acid) into the intravenous injectate of ⁶⁴Cu-RaftRGD caused 30%–50% reduction in mouse renal radioactivity levels (20, 25). However, such addition to the intraperitoneal injectate of ⁶⁴Cu-RaftRGD produced only a 19.6% reduction (obtained from a preliminary study shown in Supplementary Fig. S18A–S18C). An optimization of the dosing regimens of these renoprotective agents is hence, required.

In conclusion, intraperitoneally injected ⁶⁴Cu-RaftRGD exhibits excellent tumor penetration capability and can be specifically targeted to all forms of OCPM, including ascites cancer cells, tiny avascular lesions, and vascularized tumors of varying size. It is applicable for PET visualization and quantification of α_Vβ₃-positive OCPM and shows promise for α_Vβ₃-TRT of OCPM with a low potential risk of radiotoxicity. It is hence, a promising all-in-one radiotheranostic agent for OCPM management and worth further study. The spatially complementary ITD patterns of ⁶⁴Cu-RaftRGD and hypoxia suggest that pairing α_Vβ₃- and hypoxia-targeted TRT is a possible approach to tackle the problem of heterogeneous delivery of cytocidal radiation that is encountered with single TRT agents. Importantly, such combined TRT is also suitable for sub-millimeter sized peritoneal tumor deposits that can contain significant hypoxic fractions as reported previously (45) and shown in Supplementary Fig. S17B–S17D. In the future, when the use of α-emitting radionuclides becomes practical, incorporating these agents in a paired TRT strategy would greatly improve treatment effects, as the high linear energy transfer of α particles induces highly potent cell killing while overcoming the limited tissue pathlength of α particles, for example, 40–100 μm for α particles versus 950–1,400 μm for ⁶⁴Cu (46, 47). ²²⁵Ac and its decay product, ²¹³Bi, are pure α-emitters of interest in ovarian cancer therapy (48, 49). Given the facts that ²²⁵Ac can be stably attached to a cRGD peptide by using a bifunctional chelator, DOTA (50), and RaftRGD can be conjugated with DOTA (19), the radiolabeling of RaftRGD with ²²⁵Ac would be feasible.

A.B. Tsuji reports grants from Japan Society for the Promotion of Science during the conduct of the study, as well as grants from Japan Society for the Promotion of Science, Chugai Pharmaceutical, and Japan Agency for Medical Research and Development and personal fees from RIKEN outside the submitted work. No potential conflicts of interest were disclosed by the other authors.

Z.-H. Jin: Conceptualization, data curation, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, writing-review and editing. A.B. Tsuji: Conceptualization, data curation, supervision, investigation, writing-review and editing. M. Degardin: Investigation, methodology, writing-review and editing. A. Sugyo: Investigation, visualization. S. Obara: Software, formal analysis, methodology. H. Wakizaka: Data curation, investigation, methodology. K. Nagatsu: Investigation, methodology. K. Hu: Investigation, visualization. M.-R. Zhang: Investigation, methodology. P. Dumy: Methodology, writing-review and editing. D. Boturyn: Methodology, writing-review and editing. T. Higashi: Supervision, project administration, writing-review and editing.

This work was supported, in part, by JSPS KAKENHI grant no., JP18K07776 (Japan) and the Agence Nationale de la Recherche (ANR-17-EURE-0003, France). The authors M. Degardin and D. Boturyn wish to acknowledge support from the ICMG Chemistry Nanobio Platform (Grenoble, France), on which the peptide synthesis was performed, and the Labex ARCANE and CBH-EUR-GS (ANR-17-EURE-0003). The authors thank the Department of Advanced Nuclear Medicine Sciences and the Cyclotron Operation Section of National Institute of Radiological Sciences (Chiba, Japan) for supplying ⁶⁴Cu. We are grateful to Dr. J.-F. Gourvest (Aventis) for kindly providing HEK293(β₁) and HEK293(β₃) cells, and to Dr. J.-L. Coll (INSERM U1209, Grenoble, France) for early effort of providing IGR-OV1 cells.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.