A Targeted Thorium-227 Conjugate Demonstrates Efficacy in Preclinical Models of Acquired Drug Resistance and Combination Potential with Chemotherapeutics and Antiangiogenic Therapies

This article is featured in Selected Articles from This Issue, p. 997

Targeted thorium-227 conjugates (TTC) are an innovative class of targeted alpha therapies (TAT) for the treatment of cancer (1). TTCs consist of an antigen-targeting moiety covalently attached to a 3,2-hydroxypyridinone (HOPO) chelator, which enables stable complexation and delivery of the potent radionuclide thorium-227 to cancer cells upon systemic administration. The mode-of-action of TTCs relies on the targeted delivery of alpha particles emitted by thorium-227 that emits high linear energy transfer radiation, thereby inducing difficult-to-repair clustered DNA double-strand breaks (DSB) and forcing cells into apoptotic and necrotic cell death (2). TTCs have shown potent preclinical in vivo activity in monotherapy and in combination with DNA damage inhibitors or antiandrogen therapies (3–6).

Despite significant progress in drug development, evolving mechanisms of drug resistance remain a major hurdle for various treatment classes used for cancer therapy. A meta-analysis of survival data from numerous countries concluded in 2019 that the 5-year overall survival from ovarian cancer had improved only moderately since 1980 (https://seer.cancer.gov/statistics-network/explorer/). Approximately 314,000 cases were diagnosed in 2020 and more than 65% of the cases succumb to the disease, making ovarian cancer the eighth deadliest disease among women (7). Clinical trials using immuno-oncology (IO) approaches have shown limited efficacy in ovarian cancer, potentially due to an immunosuppressive microenvironment inherent to the disease (8). However, to date, no cellular mechanisms of resistance have been described for TAT (9).

Currently, the first-line treatment for patients with high-grade epithelial ovarian cancer is a debulking surgery (10, 11) followed by chemotherapy with carboplatin in combination with paclitaxel or docetaxel. Certain targeted therapies, including the antiangiogenic agent bevacizumab (12) and oral PARP inhibitors (13), have been approved for the treatment of ovarian cancer based on the results of randomized clinical trials showing significant benefits in terms of progression-free survival, with acceptable tolerability and no detrimental effects on quality of life. Despite this progress, most patients with ovarian cancer will still eventually experience disease recurrence, partially due to the development of cellular resistance mechanisms, including the upregulation of acquired drug resistance (ADR) markers such as P-glycoprotein (p-gp), known to decrease overall survival rates in patients with ovarian cancer (14, 15).

Mesothelin (MSLN) is a glycosylphosphatidylinositol-linked, membrane-bound cell surface protein with high and specific expression in patients with mesothelioma and ovarian cancer (16–18). MSLN-TTC is a TAT consisting of the MSLN-specific antibody anetumab and has demonstrated potent in vivo activity in cell line and patient-derived xenograft (PDX) models in monotherapy (2, 3). Furthermore, the combination of MSLN-TTC with DNA damage repair inhibitors such as PARP inhibitors and inhibitors of the Ataxia telangiectasia and Rad3 related protein (ATR) has resulted in increased efficacy compared with monotherapies in in vitro and in vivo models of ovarian cancer (4).

Because of the main mode-of-action of TTCs, their efficacy should not be hampered by cellular resistance mechanisms including expression of p-gp (9). Therefore, and based on the encouraging results with monotherapy (2, 3), we evaluated the efficacy and combination potential of the MSLN-TTC in vitro and in vivo in preclinical models of drug resistant ovarian and cervical cancers. We used the isogenic human ovarian cancer cell line pair OVCAR-8 (p-gp negative) and NCI-ADR-RES (p-gp positive) and also evaluated MSLN-TTC in vitro and in vivo in a p-gp–positive ovarian cancer PDX model in combination with various chemotherapeutics and the antiangiogenic therapies bevacizumab (approved by the FDA for the treatment of ovarian cancer and various other cancers; ref. 19) and regorafenib (FDA-approved for the treatment of advanced colorectal cancer, gastrointestinal stromal tumors, and hepatocellular carcinoma; refs. 20–23). We also examined MSLN-TTC in a model of cervical cancer where MSLN has recently been suggested as a target (24, 25).

The MSLN-targeting antibody anetumab, the MSLN-TTC containing the anetumab antibody covalently linked to a 3,2-HOPO chelator radiolabeled with thorium-227, a radiolabeled isotype control antibody, the nonradiolabeled MSLN antibody–chelator conjugate, and the radionuclide thorium-227 were manufactured by Bayer AG and Bayer AS as described previously (3, 26). Cisplatin (Teva Pharmaceuticals), doxorubicin (Sigma-Aldrich), vinorelbine (Actavis Generics), pegylated liposomal doxorubicin (Doxil, Janssen Pharmaceuticals), docetaxel (Accord Healthcare), bevacizumab (Genentech/Roche), paclitaxel (Sigma-Aldrich), regorafenib (Bayer AG), and DM4 (ravtansine, Bayer AG) were used as reference compounds.

OVCAR-3 ovarian cancer cells and the isogenic ovarian cancer cell line pair OVCAR-8 (p-gp-negative) and NCI-ADR-RES (p-gp-positive) were obtained from ATCC between 2008 and 2019. A derivative of the human cervical cancer cell line, HeLa-MaTu, expressing the p-gp drug efflux pump and therefore designated Hela-MaTu-ADR to signify the drug resistance, was obtained from EPO Berlin-Buch GmbH. The cells were authenticated using short tandem repeat DNA fingerprinting at the Leibniz Institute (German Collection of Micro-organisms and Cell Cultures, DSMZ) before the experiments and tested for Mycoplasma.

The human ovarian cancer biopsies, 24 matching primary and 24 recurring samples from 24 patients, were obtained from Provitro GmbH. Written informed consent was obtained from donors. The studies were conducted according to the Declaration of Helsinki and approved by an institutional review board.

The ovarian cancer PDX model ST206B was obtained from South Texas Accelerated Research Therapeutics. All mouse experiments were performed under the respective national animal welfare laws in Denmark, Germany, and Norway and approved by the local authorities [National Animal Experiments Inspectorare under the Ministry of Environment and Food of Denmark, Landesamt für Gesundheit und Soziales (LaGeSo), and Norwegian National Animal Research Authority, respectively].

The cytotoxicity of MSLN-TTC, paclitaxel, docetaxel, doxorubicin, cisplatin, and DM4 was determined in OVCAR-8 and NCI-ADR-RES cells. The cells were exposed to the compounds for 5 days at decreasing concentrations: for MSLN-TTC and a radiolabeled isotype control antibody starting at 20 kBq/mL and for other compounds starting at 30 μmol/L. Cell viability was analyzed using CellTiter-Glo (Promega) as previously described (3).

To determine the synergistic in vitro activity of MSLN-TTC in combination with doxorubicin and docetaxel, OVCAR-8 cells were seeded in 384-well plates (30 μL per well, 30,000 cells/mL) and after a 24-hour incubation, treated with MSLN-TTC and either doxorubicin or docetaxel using a D300e digital dispenser (Tecan). The compounds were dispensed as titrations in the following combination ratios of MSLN-TTC (C1) and doxorubicin (C2): 1 × C1, 0.8 × C1 + 0.2 × C2, 0.6 × C1 + 0.4 × C2, 0.5 × C1 + 0.5 × C2, 0.4 × C1 + 0.6 × C2, 0.8 × C1 + 0.2 × C2, and 1 × C2. MSLN-TTC was titrated in the range of 0.001–20 kBq/mL at a specific activity of 40 kBq ²²⁷Th/μg. Doxorubicin and docetaxel were titrated in the range of 0.6–30 μmol/L. The absolute IC₅₀ values were determined by using GraphPad Prism software version 7.0. The IC₅₀ isobolograms were generated by plotting the IC₅₀ values of MSLN-TTC and doxorubicin along the x and y axis, respectively. The combination index (CI) was determined according to the median-effect model of Chou–Talalay (27) with CI < 0.8 defined as synergistic effect, 0.8 ≤ CI ≤ 1.2 defined as additive effect and CI > 1.2 defined as antagonistic effect.

To study the in vitro mode-of-action via Western blot analysis, OVCAR-8 and NCI-ADR-RES cells were incubated in vitro with MSLN-TTC at 0.5 or 5 kBq/mL either alone or in combination with doxorubicin (10 nmol/L) or docetaxel (2 nmol/L) for 24 hours. The medium was removed from the cells and 1 mL of RIPA buffer was added. Cells were harvested, lysed, and protein concentrations were determined using the BCA Protein Assay Kit (Promega). Western blot analysis was performed using primary antibodies specific for the DNA damage marker phospho-H2AX (γH2AX, #9718, Cell Signaling Technology) and cell death marker cleaved PARP (#9541, Cell Signaling Technology). Ten micrograms per sample were separated on 4%–12% BisTris NuPage gel (Invitrogen) and transferred to a Western blot membrane using the iBLOT system (Invitrogen). HSP90 (anti-HSP90, BD Biosciences) was used as control. Horseradish peroxidase (HRP)-labeled secondary antibodies and DAB substrate were used for visualization.

The in vivo efficacy of MSLN-TTC was examined in cell-derived xenograft models of ovarian cancer (OVCAR-8 and NCI-ADR-RES) and human cervical cancer (HeLa-MaTu-ADR). The efficacy of MSLN-TTC as monotherapy and in combination with docetaxel, pegylated liposomal doxorubicin, bevacizumab, and regorafenib was assessed in the ST206B model. This model was also used for biodistribution studies. The efficacy of regorafenib monotherapy was assessed in the OVCAR-3 ovarian cancer model.

Mice were predosed with an irrelevant mIgG2a antibody (28) one day prior to MSLN-TTC treatments, and MSLN-TTC was used at a total antibody dose of 0.14 mg/kg in all studies with cell line–derived models and at 0.43 mg/kg in the ST206B PDX study.

In the OVCAR-8 model, 2 × 10⁶ cells were inoculated subcutaneously into female NMRI nude (nu/nu) mice (6 weeks, 22 g, Janvier Labs). Starting at an average tumor size of 123 mm³, the mice (n = 10/group) were treated intravenously with vehicle (0.9% NaCl), MSLN-TTC, or the radiolabeled isotype control antibody (both at 2 × 250 kBq/kg or 2 × 500 kBq/kg at 4-week intervals).

In the NCI-ADR-RES model, 2 × 10⁶ cells were inoculated subcutaneously into female NMRI nude (nu/nu) mice (6 weeks, 22 g, Janvier Labs). Starting at an average tumor size of 88 mm³, the mice (n = 10/group) were treated with vehicle (0.9% NaCl), MSLN-TTC or the radiolabeled isotype control antibody (both at 2 × 250 kBq/kg or 2 × 500 kBq/kg, i.v., at one-week intervals), or cisplatin [6 mg/kg, once every 7 days (Q7D), i.v.] or doxorubicin (10 mg/kg, Q7D, i.v.).

In the HeLa-MaTu-ADR model, 2 × 10⁶ cells were inoculated subcutaneously into female HspCpb athymic nude (Foxn1^nu) mice (6 weeks, 25 g, Envigo). Two separate studies were performed. In the first one, starting at an average tumor size of 151 mm³, the mice (n = 10/group) were treated with vehicle (0.9% NaCl), a single dose of nonradiolabeled MSLN antibody–chelator conjugate (0.14 mg/kg, i.v.) a single dose of MSLN-TTC or the radiolabeled isotype control antibody (both at 375 or 750 kBq/kg, i.v.), doxorubicin (10 mg/kg, Q7D, i.v.) or paclitaxel (24 mg/kg, Q7D, i.v.). In the second study, starting at the average tumor size of 110 mm³, the mice (n = 10/group) were treated with a single dose of nonradiolabeled MSLN antibody–chelator conjugate (0.14 mg/kg, i.v.), MSLN-TTC (100, 250, or 500 kBq/kg, i.v.), or the radiolabeled isotype control antibody (250 kBq/kg, i.v.), or vinorelbine (10 mg/kg, Q7Dx4, i.v.)

Female NMRI nu/nu mice (n = 10/group) were inoculated subcutaneously with 1×10⁷ OVCAR-3 cells to assess the in vivo efficacy of regorafenib. Starting at an average tumor size of 99 mm³, the mice were treated with vehicle (propylene glycol/PEG400/Pluronic), paclitaxel [8 mg/kg, once daily for 5 days (QDx5), i.v.], or regorafenib [10 mg/kg, once daily for 39 days (QDx39), p.o.].

In the ST206B PDX model, tumor fragments (5×5×5 mm) were implanted subcutaneously in NMRI nu/nu mice (RjOrI:NMRI-Foxn1^nu/Foxn1^nu; 6 weeks, 27 g, Janvier Labs).

In the ST206B monotherapy efficacy study, the mice (n = 10/group) were treated with vehicle (30 mmol/L citrate, 70 mmol/L NaCl, 0.5 mg/mL PABA, 2 mmol/L EDTA, pH 5.5, supplemented with 0.1 mg/mL IgG2a-κ antibody from murine myeloma (Sigma-Aldrich)), MSLN-TTC (1×125, 1×250, 1×500, or 2×250 kBq/kg, Q7D, i.v.), docetaxel [15 mg/kg, every 7 days for 6 times (Q7Dx6), i.v.], pegylated liposomal doxorubicin (10 mg/kg, Q7Dx6, i.v.), bevacizumab [5 mg/kg, every 5 days for 9 times (Q5Dx9), i.v.], or orally with regorafenib (30 mg/kg, QDx41) starting at the average tumor size of 134 mm³.

In the ST206B combination therapy efficacy study, the mice (n = 10/group) were treated with vehicle [30 mmol/L citrate, 70 mmol/L NaCl, 0.5 mg/mL PABA, 2 mmol/L EDTA, pH 5.5, supplemented with 0.1 mg/mL IgG2a-κ antibody from murine myeloma (Sigma-Aldrich)], nonradiolabeled MSLN antibody–chelator conjugate antibody (a single dose, 0.43 mg/kg, i.v.), MSLN-TTC (1×250 kBq/kg, i.v.), docetaxel (7.5 mg/kg, Q7Dx9, i.v.), pegylated liposomal doxorubicin (5 mg/kg, Q7Dx9, i.v.), bevacizumab (5 mg/kg, Q5Dx13, i.v.), or regorafenib (15 mg/kg, QDx63, orally) as monotherapy or in combination with MSLN-TTC starting at the average tumor size of 191 mm³. Blood samples were collected 14, 30, 49, and 64 days after the first dosing for hematologic toxicity analyses. The plasma levels of fms-related tyrosine kinase 3 ligand (FLT3L) were determined with the FLT-3 L Mouse ELISA Kit (Thermo Fisher Scientific). Tumors were isolated at the end of the study for subsequent mode-of-action analyses. Responses to treatment were determined according to the RECIST criteria (29).

In the biodistribution study with the ST206B model, the mice (n = 3/group per time point) were treated with MSLN-TTC (1×250 or 1×500 kBq/kg), the radiolabeled isotype control (1×500 kBq/kg), bevacizumab (5 mg/kg, Q5D, i.v.), or regorafenib (30 mg/kg, QD, p.o.) as monotherapy or in combination with MSLN-TTC. Tumors, liver, femur, spleen, kidneys, and blood were collected 24, 168, 336, and 504 hours after the first dosing and the remaining thorium-227 activity was determined with a high-purity germanium (HPGe) detector and converted into percentage of injected dose per gram (%ID/g).

Tumors were measured twice a week with calipers and their volume calculated with the formula: (length × width²)/2 for cell line–derived models, or 0.52 × (length × width²) for the PDX models.

The expression of p-gp and MSLN was analyzed in formalin-fixed, paraffin-embedded (FFPE) human ovarian cancer biopsies (Provitro) and on FFPE sections of OVCAR-3, OVCAR-8, NCI-ADR-RES, HeLa-MaTu-ADR, and ST206B xenograft tumors by IHC. MSLN expression was determined using the MSLN-specific antibody SP74 (Spring Bioscience) as described previously (3). P-gp expression was determined using an anti–p-gp rabbit anti-human antibody (clone EPR10364–57, Abcam). After antigen retrieval, the sections were washed with Tris buffered saline (TBS) before incubation (12 minutes, room temperature) with DAKO peroxidase blocking solution (Dako). The sections were subsequently washed, incubated (12 minutes, room temperature) with DAKO protein blocking solution followed by incubation (65 minutes, room temperature) with the anti–p-gp rabbit anti-human antibody (1:500 dilution, clone EPR10364–57, Abcam). After washing, the primary antibody was detected using HRP-labeled anti-rabbit antibody (Envision, Dako) and subsequently visualized using DAB solution.

For mode-of-action studies, the expression of MSLN, cleaved caspase-3, and γH2AX were determined in isolated NCI-ADR-RES tumors using antibodies for MSLN (SP74, Spring Biosciences), cleaved caspase-3 (Cell Signaling Technology), and γH2AX (Millipore) as previously described (3). The general histology was examined using hematoxylin & eosin (H&E) staining. For quantitative analysis of γH2AX, proteins were extracted from FFPE samples of NCI-ADR-RES tumors using the Qproteome FFPE tissue kit (Qiagen) and analyzed using western blotting. Total H2AX was detected with CST #2595 and γH2AX with CST #9718 (both Cell Signaling Technology) on a Peggy Sue instrument (ProteinSimple). The γH2AX signal was normalized relative to the total H2AX signal.

For further mode-of-action studies, ST206B PDX tumors were stained with antibodies for γH2AX (Millipore), cell proliferation marker Ki67 (Abcam), endothelial cell marker CD31 (Abcam), and smooth muscle cell marker α-SMA (Abcam). Samples were further stained with H&E and the necrotic area was determined by histology. Fluorescent visualization was performed upon incubation with the Opal Fluo kit (Akoya Bio). Quantitation was performed using the HS Analysis Webkit software (version 1.3.21, HS Analysis GmbH). Cells positive for CD31 and α-SMA were counted manually from tissue sections.

Statistical analyses were performed as indicated in Tables and Figure legends and using the statistical programming language R (version 3.5.2).

The data generated in this study are available upon reasonable request from the corresponding author.

It has been shown that the expression of the drug efflux pump p-gp correlates negatively with overall survival in patients with ovarian cancer (14). We observed p-gp expression in 7 of 48 primary and relapsing specimens of human ovarian cancer biopsies (Supplementary Fig. S1), accounting for approximately 15% of examined patient samples and aligning with the reported rate of p-gp expression in patients with ovarian cancer (30).

To identify candidate models for the evaluation of MSLN-TTC in the context of acquired drug resistance, tumor sections of the isogenic p-gp ovarian cancer models OVCAR-8 and NCI-ADR-RES were analyzed for MSLN and p-gp expression. Both showed MSLN expression, but the level was slightly higher in OVCAR-8 cells than in NCI-ADR-RES cells (Supplementary Table S1). P-gp was confirmed to be expressed in NCI-ADR-RES tumors but not in OVCAR-8 tumors (Supplementary Fig. S2). Tumors of a third xenograft model, the human cervical cancer model Hela-MaTu-ADR with acquired drug resistance, showed weak expression of MSLN (Supplementary Table S1) and were determined as p-gp–positive. Finally, expression of MSLN and p-gp was demonstrated in the OvCa PDX model ST206B originating from a patient with metastatic high-grade papillary serous ovarian carcinoma who had been on several treatment lines, including paclitaxel, carboplatin, gemcitabine, and topotecan (Supplementary Fig. S2).

The efficacy of MSLN-TTC was assessed in in vitro cytotoxicity assays using OVCAR-8 and NCI-ADR-RES cells and compared to the standard-of-care therapeutics paclitaxel, docetaxel, doxorubicin, and cisplatin. In addition, the maytansinoid DM4, a toxophore used in antibody–drug conjugate approaches, was included as a reference compound (31). The potency of MSLN-TTC was almost identical in both cell lines (IC₅₀ 5.2×10⁻¹² and 1.2×10⁻¹² on OVCAR-8 and NCI-ADR-RES, respectively) and its cytotoxicity was specific, as evidenced by the radiolabeled isotype control demonstrating an approximately 10-fold lower IC₅₀ value in both cell lines. In contrast, all chemotherapeutics except for cisplatin showed decreased potency in p-gp–positive NCI-ADR-RES cells compared with p-gp–negative OVCAR-8 cells, however, to different degrees (Supplementary Table S2). Both taxanes were 100-fold less potent in p-gp–positive NCI-ADR-RES cells than in p-gp–negative OVCAR-8 cells, whereas doxorubicin and DM4 were approximately 40-fold and 17-fold less potent, respectively.

As doxorubicin and docetaxel are approved first-line treatments in ovarian cancer, we investigated their combination potential with MSLN-TTC. In OVCAR-8 cells, both combinations showed additive cytotoxic activity with a calculated combination index of 0.9 (Supplementary Fig. S3A and S3B; Supplementary Table S3). We also investigated the underlying molecular mechanisms of the additive effect observed with combination treatments. OVCAR-8 cells exposed to MSLN-TTC alone at activities of 0.5 and 5 kBq/mL showed a dose-dependent increase in the DNA damage marker γH2AX in comparison to untreated cells (Supplementary Fig. S3C and S3D), which is in line with previously published data for MSLN-TTC (3). The combination treatments with either doxorubicin or docetaxel showed further increase in γH2AX compared to all three monotherapies. Similarly, compared with the three monotherapies, increased levels of the cell death marker cleaved PARP were also detected when MSLN-TTC was combined with doxorubicin at 5 kBq/mL and with docetaxel at 0.5 kBq/mL (Supplementary Fig. S3C and S3D).

On p-gp–positive NCI-ADR-RES cells, MSLN-TTC cytotoxicity was enhanced when it was combined with doxorubicin, but not with docetaxel (Supplementary Fig. S3E and S3F; Supplementary Table S3). This is in line with the finding of doxorubicin monotherapy being more potent than paclitaxel in NCI-ADR-RES cells (Supplementary Table S2). On the molecular level, no further increase of γH2AX and cleaved PARP was observed when MSLN-TTC was combined with either doxorubicin or docetaxel (Supplementary Fig. S3G and S3H).

In summary, the efficacy of MSLN-TTC is not influenced by the drug efflux pump p-gp and it shows combination potential with both DNA and microtubule-interfering agents in p-gp–negative models in vitro.

The antitumor efficacy of MSLN-TTC was evaluated in vivo using the cell line-derived OVCAR-8, NCI-ADR-RES, and HeLa-MaTu-ADR xenograft models in mice.

MSLN-TTC inhibited the growth of OVCAR-8 and NCI-ADR-RES tumors and to the same extent, indicating that the activity is independent of p-gp expression (Fig. 1A and B). MSLN-TTC showed almost equal antitumor activity at 2×500 kBq/kg and at 2×250 kBq/kg in both models. Furthermore, the effect was specific, as the radiolabeled isotype control did not show activity (Supplementary Table S4). Compared to cisplatin and doxorubicin, MSLN-TTC showed higher efficacy in the NCI-ADR-RES model, as indicated by its higher treatment/control (T/C) ratio value compared to the reference compounds (Supplementary Table S4). IHC analyses of NCI-ADR-RES tumor samples demonstrated a dose-dependent induction of DNA damage as indicated by the increasing level of γH2AX in the MSLN-TTC-treated tumors compared to the vehicle group (Supplementary Fig. S4A). Increased levels of the apoptosis marker cleaved caspase-3 and a marked increase in necrotic areas suggest additional, MSLN-TTC-induced damage in NCI-ADR-RES tumors leading to the infiltration of macrophages and monocytes into the tumor (Supplementary Fig. S4B).

The efficacy of MSLN-TTC was further evaluated in two studies performed with the HeLa-MaTu-ADR cervical cancer model. In the first study, MSLN-TTC was compared to pegylated liposomal doxorubicin and paclitaxel. As MSLN is also highly expressed in mesothelioma, vinorelbine, the current standard-of-care treatment for mesothelioma, was included for comparison in the second study (32). MSLN-TTC demonstrated potent, dose-dependent efficacy when administered at single doses of 375 or 750 kBq/kg (Fig. 1C; Supplementary Table S4) and both doses were well tolerated (Supplementary Fig. S5A). The efficacy of MSLN-TTC was specific, as a radiolabeled isotype control did not show any tumor growth delay at the same dose level. Of the reference compounds, pegylated liposomal doxorubicin showed antitumor activity whereas paclitaxel and vinorelbine did not (Fig. 1C; Supplementary Fig. S5B; Supplementary Table S4).

In summary, MSLN-TTC demonstrated marked and selective in vivo efficacy irrespective of the p-gp status of the model and was well tolerated without significant animal weight loss.

The efficacy of MSLN-TTC was further evaluated in several studies using the human MSLN- and p-gp–positive ovarian cancer PDX model ST206B.

In the first study, the efficacy of MSLN-TTC was investigated as monotherapy and compared to the chemotherapies pegylated liposomal doxorubicin (10 mg/kg, Q7Dx6) and docetaxel (15 mg/kg, Q7Dx6) as well as the angiogenic inhibitor bevacizumab (5 mg/kg, Q5Dx9) and the multi-kinase inhibitor regorafenib (30 mg/kg, QDx41), both described to block VEGF receptor signaling (20, 33). Regorafenib alone (10 mg/kg, QDx39) had shown marked antitumor efficacy in the MSLN- and p-gp–positive OVCAR-3 model (34) with a T/C value of 0.21, supporting its combination potential with MSLN-TTC (Supplementary Fig. S6). MSLN-TTC was applied at single doses of 125, 250, 500 kBq/kg, or at 2×250 kBq/kg at one-week intervals. Specific and dose-dependent in vivo potency was observed for MSLN-TTC when applied at doses of 125, 250, 2×250 or 1×500 kBq/kg with T/C values of 0.58, 0.24, 0.11, and 0.15, respectively, on day 40 (Table 1; Supplementary Fig. S7A and S7B). The reference compounds docetaxel, pegylated liposomal doxorubicin, regorafenib, and bevacizumab also showed clear tumor growth inhibition as monotherapy with T/C values ranging from 0.17 to 0.42, while the radiolabeled isotype control (1×500 kBq/kg) was not efficacious.

In the second ST206B study, the efficacy of MSLN-TTC was further evaluated in combination with docetaxel, pegylated liposomal doxorubicin, regorafenib, and bevacizumab in the ST206B model (Fig. 2). Compared to the ST206B monotherapy study described above, lower doses were chosen to demonstrate a differential effect between mono- and combination therapy. A single dose of 250 kBq/kg MSLN-TTC was now combined with bevacizumab (5 mg/kg, Q5Dx13) or half-doses of pegylated liposomal doxorubicin (5 mg/kg, Q7Dx9), docetaxel (7.5 mg/kg, Q7Dx9), or regorafenib (15 mg/kg, QDx63) compared to the monotherapy study. The combination treatments resulted in a stronger tumor growth inhibition as evidenced by the lower T/C ratios (Fig. 2A–D; Table 1) and increased numbers of responses (Fig. 2E; Supplementary Table S5) compared to the monotherapies. The combination effect was determined to be additive for all four combination treatments. Tumor weights were lower in the MSLN-TTC and docetaxel combination treatment group compared with the docetaxel monotherapy group (Fig. 2F).

The biodistribution of MSLN-TTC was evaluated in the ST206B model and MSLN-TTC showed specific accumulation in the MSLN-expressing tumor over the course of three weeks (504 h) after dosing, reaching approximately 50% ID/g (Supplementary Fig. S8). In comparison, a radiolabeled isotype control showed only minor accumulation of approximately 5–10% ID/g, which can be attributed to the enhanced penetration and retention effect of large biological tumors (35). To assess the impact of angiogenesis inhibitors on the tumor accumulation of MSLN-TTC, a parallel arm of the study examined the biodistribution of MSLN-TTC in the presence of bevacizumab or regorafenib (Fig. 3A). Tumor-specific accumulation of MSLN-TTC was observed in both monotherapy and combination groups, with the highest activity determined at 336 hours (Fig. 3A). Although there was a trend of a 27–34% lower accumulation in the groups where MSLN-TTC was combined with bevacizumab or regorafenib compared with the MSLN-TTC monotherapy group, the differences between the treatment groups did not reach statistical significance. No difference in thorium-227 accumulation was observed in any other organs between the monotherapy and combination treatments.

Next, modes-of-action such as necrosis, DNA damage, proliferation, and angiogenesis were analyzed in tumor samples from the ST206B combination study 64 days after start of the treatment. None of the treatments increased necrosis (Fig. 3B). MSLN-TTC, docetaxel and pegylated liposomal doxorubicin monotherapies showed a trend of slight increase in the DNA damage marker γH2AX compared with vehicle, but no additive increase was observed in the respective combination groups (Fig. 3C). Significant decrease of the proliferation marker Ki67 was detected in the regorafenib monotherapy group, and in all combination treatment groups compared with vehicle, with the strongest effect observed for the combination of MSLN-TTC and regorafenib (Fig. 3D). Due to the antiangiogenic properties of regorafenib and bevacizumab, the number of CD31/α-SMA double-positive blood vessels in tumors was also analyzed (Figs. 3E and 4; Supplementary Fig. S9). A significant decrease in the number of double-positive vessels was observed in all combination treatment groups compared with vehicle. A significant decrease was also observed with regorafenib, bevacizumab, and interestingly, MSLN-TTC monotherapies compared with vehicle.

All monotherapy and combination treatments with MSLN-TTC and the reference compounds were well-tolerated in the ST206B combination study without significant animal weight loss (Fig. 5A). In the same study, a suppression of white blood cells was observed for MSLN-TTC and pegylated liposomal doxorubicin monotherapies and all four combination treatment regimens, but the effect was reversible (Fig. 5B–E). A temporary suppression of red blood cells was observed with the combination of MSLN-TTC with the antiangiogenic therapies regorafenib and bevacizumab (Supplementary Fig. S10). An approximately 2–3-fold increase from baseline was observed in the levels of FLT3L, a growth factor known to activate hematopoietic cells, in mice treated with MSLN-TTC either alone or in combination (Fig. 5F–I). When pegylated liposomal doxorubicin or regorafenib were combined with MSLN-TTC, the treatment resulted in further increase in FLT3L levels compared with MSLN-TTC monotherapy 14 or 30 days after treatment, but the same was not observed for the MSLN-TTC combinations with bevacizumab or docetaxel.

Despite undoubtable progress in the development of anticancer drugs especially in the field of IO in the recent years, there is still an unmet medical need in the treatment of solid cancers such as high-grade serous ovarian cancer. In ovarian cancer, treatment options are confined to surgery followed by combination chemotherapy with most patients relapsing after therapy. This is often due to the codevelopment of resistance mechanisms including the expression of p-gp, which is known to neutralize the therapeutic effect of several chemotherapeutics, resulting in decreased clinical benefit (14). In this study, p-gp expression was observed in approximately 15% of primary and relapsing patient samples, confirming the previously published data (30). As the development of resistance mechanisms commonly leads to relapse, new treatment options with different modes-of-action are clearly required. Here, based on the previously published encouraging findings, we investigated the potential of the TAT MSLN-TTC (3) as a single treatment and in combination with first-line standard-of-care treatments in both in vitro and in vivo models of drug resistant ovarian and cervical cancer and also in models with a negative p-gp status. Compared with, for example, antibody–drug conjugates, the mode-of-action of TTCs does not require internalization nor does the radionuclide (= toxophore) need to be enzymatically released in the cytoplasm, which renders TTCs and TATs in general less susceptible to development of acquired drug resistance.

MSLN-TTC demonstrated undiscriminating in vitro potency regardless of the p-gp status, both in the p-gp–negative OVCAR-8 cell line as well as in the p-gp–positive NCI-ADR-RES cell line. In the NCI-ADR-RES model, the efficacy of MSLN-TTC was also shown to be superior to the chemotherapeutics approved for the treatment of ovarian cancer and mesothelioma, that is, paclitaxel, doxorubicin, cisplatin, and vinorelbine (36, 37). The combination potential of MSLN-TTC with docetaxel and doxorubicin was also evaluated. In vitro, additive activity on p-gp–negative OVCAR-8 cells was observed with both combinations, as evidenced by cell viability and analysis of DNA damage and apoptotic cell death. However, the effect was diminished when the combinations were tested in p-gp-positive NCI-ADR-RES cells, presumably resulting from the reduced efficacy of the chemotherapeutics on p-gp–positive cells.

The difference in efficacy between MSLN-TTC and the tested chemotherapeutics in the p-gp–positive NCI-ADR-RES cells most likely arises from the differing modes-of-actions of the treatments in relation to the manner by which p-gp confers resistance. The p-gp is a protein located on the cell membranes of various tissues that is involved in the traffic of substrates outside the cells (38, 39), and in a considerable number of cancers its expression is one of the most commonly encountered mechanisms of multidrug resistance representing a major obstacle to successful chemotherapy (40, 41). The consequences of this mechanism were also evident in our study as paclitaxel and doxorubicin, known substrates of the p-gp efflux pump (30, 42, 43), demonstrated much lower IC₅₀ values in the p-gp–positive cell line compared to the p-gp–negative cell line, whereas cisplatin, which is not a p-gp substrate (44), only showed slightly lessened potency. The mode-of-action of MSLN-TTC, on the other hand, relies on the targeted delivery of alpha particles that induce DSBs that result in apoptotic and necrotic cell death (2), explaining its strong efficacy regardless of the p-gp status.

The efficacy of TAT by MSLN-TTC was also demonstrated to be present in vivo both as a single treatment and in combination with chemotherapeutics. Much like in the in vitro models, as monotherapy, MSLN-TTC demonstrated marked and selective in vivo efficacy, irrespective of the used model's p-gp status. The efficacy was dose-dependent and higher compared to the used reference compounds cisplatin, doxorubicin, paclitaxel, and vinorelbine. MSLN-TTC combination efficacy was studied in the ovarian cancer PDX model ST206B which originates from a patient treated with several lines of chemotherapeutics including paclitaxel, carboplatin, and gemcitabine and shows strong MSLN and weak p-gp expression. Here, MSLN-TTC as well as all the four reference compounds demonstrated strong antitumor efficacy when administered as single agents, but importantly, this efficacy was further enhanced in an additive manner when a sub-efficacious dose of 1×250 kBq/kg of MSLN-TTC was combined with half of the monotherapy doses of either docetaxel, pegylated liposomal doxorubicin, or regorafenib, or the full monotherapy dose of bevacizumab. Interestingly, despite weak p-gp expression and the fact that docetaxel and pegylated liposomal doxorubicin are substrates of the p-gp pump, the combination of MSLN-TTC and docetaxel resulted in the strongest antitumor efficacy, followed by the combinations with pegylated liposomal doxorubicin and regorafenib. The combination of MSLN-TTC with bevacizumab was also superior to the respective monotherapy but showed weaker antitumor efficacy than the other combination treatments, as evidenced by the T/C ratio and the reduced number of partial responses. Another explanation for the higher synergy exhibited by the MSLN-TTC combination treatments with docetaxel or doxorubicin is the fact that both compounds act on mitotic arrest which complements the DNA damage-based mode of action of MSLN-TTC.

The differences in the efficacies of the MSLN-TTC combination therapies could have also been affected by differences in thorium-227 accumulation, in addition to variance in reaction to p-gp expression, and previous research has shown that the tumor accumulation of zirconium-89-radiolabeled trastuzumab conjugate is decreased when bevacizumab is co-administered (45). The reduction in thorium-227 accumulation is probably explained by the decrease of newly formed blood vessels in the tumors. CD31/α-SMA costaining, which indicates co-presence of endothelial cells and pericytes and thus vessel maturation (46), was demonstrated to be reduced in the groups where MSLN-TTC was combined with bevacizumab or regorafenib. This indicates that antiangiogenic inhibitors might not show the most optimal combination potential when used with large molecules such as antibodies. Regarding other mechanistic analyses, none of the treatments increased necrosis, but MSLN-TTC, docetaxel, and pegylated liposomal doxorubicin monotherapies showed a trend of slight increase in the DNA damage marker γH2AX, and a significant decrease of the proliferation marker Ki67 was detected in the regorafenib monotherapy group and all four combination treatment groups compared with vehicle.

Importantly, all combinations were well-tolerated, although suppression of blood cells was observed for MSLN-TTC and pegylated liposomal doxorubicin in monotherapy and in treatment regimens where MSLN-TTC was combined with docetaxel, pegylated liposomal doxorubicin, regorafenib, or bevacizumab. An increase of FLT3L, known to simulate the proliferation and differentiation of blood progenitor cells (47), was observed for treatment regimens comprising MSLN-TTC. This could be a reaction by the cells to the treatment and therefore, FLT3L could potentially be considered as a biomarker to study hematologic side effects.

In summary, this study demonstrates that the efficacy of MSLN-TTC is independent of p-gp status and can thus overcome mechanisms of resistance to existing therapies of ovarian cancer. Furthermore, in addition to our previous reports on MSLN-TTC in combination with immuno-oncological drugs (48) or DNA damage response inhibitors (4), we show here that MSLN-TTC demonstrates additive antitumor activity in vivo in a PDX model of ovarian cancer when combined with the chemotherapeutics docetaxel and doxorubicin or with the angiogenesis inhibitors regorafenib or bevacizumab.

S. Zitzmann-Kolbe reports personal fees from Bayer AG during the conduct of the study; personal fees from Bayer AG outside the submitted work. C. Kamfenkel reports I am a shareholder of Bayer AG. O. Politz reports personal fees from Bayer AG during the conduct of the study. M.U. Juul reports grants from Bayer AG during the conduct of the study. J. Fonslet reports other support from Bayer AG during the conduct of the study. C.H. Nielsen reports grants from Bayer AG during the conduct of the study. C.A. Schatz reports I am a full time employee of Bayer AG and I do own Bayer AG stocks. D. Mumberg reports I am an employee and a stockholder of Bayer AG. No disclosures were reported by the other authors.

S. Zitzmann-Kolbe: Resources, data curation, formal analysis, methodology, writing–original draft. A. Kristian: Resources, data curation, formal analysis, methodology, writing–original draft. D. Zopf: Conceptualization, investigation, writing–original draft, writing–review and editing. C. Kamfenkel: Resources, data curation, software, formal analysis. O. Politz: Conceptualization. C. Ellingsen: Resources, data curation, formal analysis, methodology, writing–original draft. J. Hilbig: Formal analysis, methodology. M.U. Juul: Resources, formal analysis, supervision. J. Fonslet: Resources, formal analysis, supervision. C.H. Nielsen: Resources, formal analysis, supervision. C.A. Schatz: Conceptualization, resources, software. R.M. Bjerke: Resources, supervision. A.S. Cuthbertson: Conceptualization, resources. D. Mumberg: Conceptualization, investigation. U.B. Hagemann: Conceptualization, resources, data curation, software, formal analysis, supervision, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

We would like to thank Maria Spelling, Seren Nesan, Oliver Gernetzki, Manuela Brand and Manuela Steinbach for excellent technical support. Riina-Minna Väänänen and Sanna-Maria Käkönen at Aurexel Life Sciences Ltd. (www.aurexel.com) are acknowledged for editorial support funded by Bayer AG.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.