Ovarian and uterine carcinosarcomas are sensitive in vitro and in vivo to Elimusertib, a novel ataxia-telangiectasia and Rad3-related (ATR) kinase inhibitor

Abstract

Background:

Carcinosarcoma of the ovary (OCS) and uterus (UCS) are rare highly aggressive malignancies. Ataxia-telangiectasia-and-Rad3-related (ATR) kinase and homologous recombination play a pivotal role in DNA damage repair. Homologous recombination deficiency (HRD) has been demonstrated in >30% of OCS/UCS. We investigated the preclinical activity of elimusertib, a selective ATR kinase inhibitor, against carcinosarcoma (CS) cell lines and xenografts.

Methods:

Sensitivity to elimusertib was evaluated in vitro against nine whole exome-sequenced (WES) primary CS cell lines and in vivo against HRD CS xenografts. Western blots were performed to determine baseline ATR and p-ATR protein expression in CS, and ATR pathway downstream effectors and apoptosis markers in CS HRD cell lines after Elimusertib treatment.

Results:

Out of the 9 CS cell lines, 3 harbored HRD and 6 homologous recombination proficient (HRP) features. Most of CS (i.e., 7/9 = 85%) were found to be sensitive to Elimusertib in vitro. Amongst the 5 primary CS cell lines with a high-grade pure serous epithelial component, HRD cell lines were more sensitive to elimusertib than HRP tumors (mean IC₅₀ ± SEM HRD CS=61.3 nM ±15.2 vs HRP=361.6nM ±24.4 (p=0.01)). Baseline ATR and p-ATR protein expression was higher in HRD CS cell lines. Elimusertib showed tumor growth inhibition in HRD CS xenografts (p<0.0001) and increased overall animal survival (p<0.0001). Western blot demonstrated dose-dependent inhibition of ATR, p-ATR and its downstream effector p-CHK1, and a dose-dependent increase in caspase-3 expression.

Conclusions:

Elimusertib is preclinically active in vitro and in vivo against primary CS cell lines and xenografts, respectively. CS models harboring HRD or with pure/mixed endometrioid histology demonstrated higher sensitivity to ATR inhibition. Clinical trials with elimusertib in CS patients are warranted.

1. Introduction

Carcinosarcomas (CS) of the gynecologic tract are rare metaplastic malignancies [1]. While they are most likely to originate within the uterus, they only account for less than 5% of all endometrial cancers [2]. Uterine CS (UCS) are characterized by a poor prognosis; 5-year overall survival (OS) rates are only 33% [3]. The treatment of CS consists of cytoreductive surgery followed by adjuvant chemotherapy, typically carboplatin and paclitaxel [3]. Despite advances in treatment effectiveness and survival for many other gynecologic malignancies, the average 5-year OS of patients with CS has not significantly changed over the past 50 years, highlighting the relative inadequacy of currently available cytotoxic chemotherapy regimens [1,2]. Accordingly, there remains an unmet clinical need for the development of novel, effective therapies active against UCS and ovarian CS (OCS).

Recent whole exome sequencing (WES) analyses of UCS and OCS tumors have unequivocally demonstrated that UCS and OCS are epithelial biphasic carcinomas with both carcinomatous and sarcomatous elements derived from a common precursor, and thus can have mutations typical of carcinomas [4]. In this vein, WES performed on both USC and OCS identified Signature-3 [5], a genetic signature associated with homologous recombination-based DNA double-strand break repair deficiency (HRD), in 60% of OCS and 25% of UCS [5,6]. These data suggest that HRD may play a pivotal role in the complex DNA damage repair (DDR) of CS and may represent a mechanism of tumorigenesis in at least a subset of these rare, highly biologically aggressive gynecologic tumors.

Ataxia telangiectasia and Rad3-related protein (ATR), ataxia telangiectasia mutated (ATM), and DNA dependent protein kinase (DNA-PK) play a central role in DNA repair by activating essential signaling pathways [7]. ATM and DNA-PK are primarily involved in double-strand break repair and ATR is primarily involved in DNA repair of replicative stress [5]. As such, HRD and ATRX/DAXX/ATM mutations may lead to increased dependency on ATR gene function for cancer cells’ rescue from mitotic catastrophe, making it an appealing target for inhibition in cancer treatment. While PARP inhibitors (PARPi) are now standard of care in the treatment of multiple human tumors characterized by HRD [8], recent studies have shown that cancer models carrying DNA damage repair deficiencies have increased sensitivity to ATR inhibitors (ATRi) [7,9]. Currently, there are several ATRi in Phase I-II trials that have demonstrated promise, including BAY1895344 (elimusertib), a potent highly selective inhibitor of ATR kinase activity [9]. In preclinical models of human cancers with varying DDR gene mutations (for example ATM, ATR, ATRX, BRIP1, MLH1, CHK1, ARID1A), elimusertib demonstrated significant activity both as monotherapy as well as in combination with varying agents including olaparib and rucaparib [10]. Despite these promising findings, ATRi have not been studied against carcinosarcomas. We sought to determine the in vitro and in vivo preclinical activity of elimusertib against a panel of fully characterized (i.e., WES) primary HRD and HR proficient (HRP) OCS and UCS cell lines and xenograft models.

2. Materials and Methods

2.1. Establishment of primary uterine and ovarian CS cell lines

The study protocol was approved by the Yale Human Investigation Committee. UCS and OCS tumor samples were collected from patients at the time of their primary surgical staging, and after sterile processing of fresh tumor biopsy samples, primary UCS and OCS cell lines were established as previously described [4,6]. Tissue source and characteristics of the fully sequenced primary CS cell lines are described in Table 1. Primary CS cell lines were genetically sequenced by WES at the Yale Center for Genome Analysis and cryopreserved after establishment as previously described [4]. Tumors were staged according to the International Federation of Gynecology and Obstetrics staging system (FIGO) [11,12]. All primary OCS and UCS cell lines used in the experiments described in our study had limited passages (i.e., <50).

Table 1.

Tumor sample characteristics

Code	Age	Race	FIGO stage	Tissue of origin	Histology
					EC	SC
SARARK 1	70	A	IC	Uterus	END+CC	ESS
SARARK 3	74	W	IIIC	Ovary	SER	CDRS
SARARK 4	77	W	IIIC	Ovary	SER	CDRS
SARARK 9	66	W	IIIC2	Uterus	SER	ESS
SARARK 7	55	W	IV	Ovary	CC+SER	CDRS
SARARK 11	67	W	IIIC1	Uterus	END	ESS+CDRS
SARARK 12	39	W	IVB	Uterus	SER	ESS
SARARK 13	72	W	X	Uterus	SER	n/a
SARARK 14	59	B	IVB	Uterus	END+SER	ESS

A: Asian; W: White; B: Black or African American

FIGO: International Federation of Gynecology and Obstetrics

EC: epithelial component; SC: sarcomatous component; END: endometrioid; CC: clear cell; ESS: endometrial stromal sarcoma; SER: serous; CDRS: chondrosarcoma.

2.2. Elimusertib (BAY1895344)

Elimusertib was obtained from Bayer AG through a material transfer agreement as previously described [10]. Briefly, elimusertib was prepared into a stock weekly with a vehicle of PEG 400, water, and 100% ethanol and kept at 4°C shielded from light.

2.3. Whole-exome DNA sequencing and copy number variant (CNV) analysis

Whole-exome sequencing, single nucleotide variant (SNV) and CNV analysis on UCS and OCS primary cell lines were undertaken as previously described [13].

2.4. Analysis of mutational signatures of CS

WES data from all 9 primary UCS and OCS CS cell lines were analyzed for mutational signatures as described by Alexandrov et al. [5]. Briefly, mutational signatures were extracted using base substitutions and additionally included information on the sequence context of each mutation. As there are six classes of base substitution C>A, C>G, C>T, T>A, T>C, T>G (all substitutions are referred to by the pyrimidine of the mutated Watson-Crick base pair), and information on the bases flanking each mutated base are incorporated in this analysis, there are 96 possible mutations in this classification. In published studies, applying this approach to multiple human cancer types has revealed over 30 distinct validated mutational signatures [14]. Importantly, signature 3 was strongly associated with BRCA1/2 mutations within the ovarian, breast and prostate cancer types (i.e., HRD-related) [5,6].

2.5. Cell viability assays

CS cell lines were plated in the log phase of growth in 6-well tissue culture plates at a density of 80,000–100,000 cells/well. After 24 hours of incubation, cells were treated with elimusertib at concentrations of 0, 0.01, 0.05, 0.1, 0.5, and 1μM. After 72 hours of drug treatment, cells were harvested in their entirety, centrifuged, and stained with propidium iodide (2 μL of 500 μg/ml stock solution in PBS). The viable cells were then quantified using flow-cytometry as mean ± SEM relative to untreated cells as 100% viable controls. A minimum of three independent experiments per cell line were performed.

2.6. Western blotting

Cells were cultured in RPMI 1640 medium (Life Technologies, Grand Island, NY) containing 10% fetal bovine serum (Seradigm, Radnor, PA) and penicillin/streptomycin (Life Technologies) at 37°C and 5% CO₂. They were treated with a medium containing elimusertib (0.3 μM or 3 μM) or DMSO solvent (0.03%) which functioned as the untreated control. At the desired time points, cells were trypsinized and collected for preparing protein lysates. Protein lysates were prepared with lysis buffer (1% Triton X-100, 0.05% SDS, 100 mM Na₂HPO4, and 150 mM NaCl) and then electrophoresed on a 4-20% pre-cast SDS-polyacrylamide gel (Bio-Rad) and transferred onto Amersham Hybond 0.45 PVDF membranes (GE Healthcare, Chicago, IL). After blocking with 5% non-fat milk in PBS-0.05% Tween 20, the membranes were incubated with primary antibodies at 4°C overnight, and then secondary antibodies for 1 h at room temperature. Antibodies include anti-p-ATR antibody (#2853, Cell Signaling Technology), anti-ATR antibody (#13934, Cell Signaling Technology), p-CHK1 antibody (#2348, Cell Signaling Technology), cleaved-caspase-3 antibody (#9664, Cell Signaling Technology), HRP-conjugated anti-β-actin antibody (#HRP-6008, ProteinTech), anti-rabbit secondary antibody (#7074, Cell Signaling Technology). The blots were developed using Clarity or Clarity Max Western ECL Blotting Substrates (Bio-Rad).

2.7. Caspase-3 activity assays

Caspase-3 activity was evaluated using Caspase-Glo 3/7 Assay kit (Promega). Protein lysate (10 μg) was diluted to a final volume of 50 μL. An equal volume of Caspase-Glo 3/7 Reagent was added to the lysate and incubated at room temperature for 1 h before they were recorded using a Navigator Microplate Luminometer.

2.8. Carcinosarcoma PDX model and treatment

The in vivo antitumor activity of elimusertib was tested in a xenograft CS model. Briefly, four to six-week-old CB17/SCID mice were given a single subcutaneous injection in the abdominal region of 7 × 10⁶ SARARK3 cells in approximately 300 μl of a 1:1 suspension of sterile PBS containing cells and Matrigel^® (BD Biosciences). Xenografted mice were randomized into treatment groups (7 mice per group) when mean tumor burden was 0.15-0.30 cm³, and dosing (vehicle PO or elimusertib 20 mg/kg BID, PO) were delivered to the SARARK3 xenografts for up to 4 weeks (7 days/week). Drug dosage was chosen according to previous studies [10]. Tumor and weight measurements of each mouse were recorded twice weekly. Mice were humanely euthanized when tumor volume reached 1.0 cm³ using the formula (width² × height) / 2. Animal care and euthanasia were carried out according to the rules and regulations as set forth by the Institutional Animal Care and Use Committee (IACUC).

2.9. Statistical analysis

Statistical analysis was performed using GraphPad Prism version 8 (Graph Pad Software, Inc. San Diego, CA). The inhibition of proliferation in CS cell lines after exposure to elimusertib were evaluated by two-tailed unpaired student t-test. Caspase-3 activity of cells in different treatment groups were compared using one-way ANOVA test. One-way ANOVA and unpaired t-test were used to evaluate significant differences in the tumor volumes at specific time points in the in vivo experiments. Overall survival data were analyzed and plotted using Kaplan-Meier survival curves, which were compared using the log-rank test. Differences in all comparisons were considered statistically significant at p-values < 0.05.

3. Results

3.1. Genetic signatures of carcinosarcoma cell lines

A total of 9 fully characterized primary CS cell lines were available for our preclinical experiments with elimusertib (Table 1). Six cell lines originated from the uterus while 3 originated from the ovary. Five of the CS cell lines harbored a pure high-grade serous epithelial component, three had a mixed epithelial component, and one harbored a high grade endometroid epithelial component (Table 1). When we analyzed the tumor mutational signatures of the 9 primary fully sequenced CS cell lines, 3 were found to harbor dominant HRD-related signatures (2 OCS and 1 UCS), while 6 were HR proficient (HRP), (5 UCS and 1 OCS). Within the HRP group, 5 demonstrated a dominant aging-related signature, with the remaining one demonstrating a microsatellite instable (MSI) signature (Figure 1.A). Analyses of the genetic characteristics of the primary CS cell lines including SNV and CNV mutations have recently been presented using a much larger set of fresh ovarian and uterine CS (i.e., 68 samples) [3,6]. As representatively demonstrated in Figure 1B, TP53 gene alterations were detected in most of the CS cell lines, with additional derangements detected in multiple driver genes affecting the ERBB2/PI3K/AKT/mTOR/PTEN pathway, the cell cycle, chromatin, and remodeling pathway and, importantly, the HRD and ATR/ATRX/DAXX/ATM/ARID1A pathways (Figure 1.B). In this regard, ATR and DAXXgene amplifications were found in SARARK 13, while SARARK 11 and SARARK 1 both hosted an ARID1A missense mutation (Figure 1.B).

Figure 1.

A) Tumor mutational signatures. HRD signatures are identified in 3 CS cell lines. Ageing signature is identified in 5 CS cell lines. One CS cell line (SARARK 11) has microsatellite instable signature. 1.B) Representation of mutations found on whole exome sequencing of tumor cell lines.

3.2. UCS and OCS CS cell lines show sensitivity to Elimusertib in vitro

To evaluate the potential activity of ATR inhibition in CS harboring HRD and HRP signatures, we investigated the in vitro effects of elimusertib on the growth of all 9 primary CS cell lines available using a flow cytometric-based assay. Using a previously established sensitivity cut off of 100 nM IC₅₀ [10], we found that 7 out of 9 (85%) primary CS cell lines were sensitive to ATR inhibition in vitro (Figure 2). When we compared the 5 CS cell lines with a high-grade serous component (i.e., 3 HRD vs. 2 HRP), we found that all 3 primary CS harboring a dominant HRD-related signature (i.e., signature 3) were significantly more sensitive to elimusertib than the 2 HRP tumors (mean elimusertib IC₅₀ ± SEM of the HRD vs. HRP cell lines were 61.3 ± 15.2 nM vs. 361.6 + 24.4 nM, p = 0.01, Figure 2.B). The remaining 4 HRP CS cell lines with pure endometrioid or mixed endometrioid/CC histology (SARARK 1, 7, and 14) were also sensitive to ATR inhibition (mean elimusertib IC₅₀± SEM was 40.0 ± 15.0 nM) (Figure 2).

Figure 2.

Antitumor activity of elimusertib on CS cell lines in vitro. 2.A) IC₅₀ dose-response curves of elimusertib against all CS cell lines. 2.B) Mean elimusertib IC₅₀ values in accordance with HRD status. 2.C) Mean elimusertib IC₅₀ values in each CS cell line.

3.3. ATR and p-ATR protein expression and downstream expression changes after elimusertib exposure

Next, we evaluated the protein expression levels of ATR and p-ATR in all nine primary CS cell lines exposed to elimusertib by western blot. The ATR and p-ATR protein expression levels in the HRD cell lines (SARARK3, SARARK 4, and SARARK9) were higher than those found in the HRP cell lines (SARARK1, SARARK 7, SARARK11, SARARK12, SARARK 13, and SARARK14) (Figure 3.A). When downstream expression changes were evaluated after elimusertib exposure in all nine primary CS cell lines, as representatively demonstrated in Figure 3.B for SARARK 3, elimusertib treatment for 0.5 h and 1 h significantly reduced the level of phosphorylated ATR (p-ATR). Since checkpoint kinase CHK1 is known to undergo ATR-mediated phosphorylation and activation we also evaluated its expression levels. As depicted in Figure 3.B we found the inhibition of ATR by elimusertib treatment to also lead to a reduction of phosphorylated CHK1 (p-CHK1). Finally, we found a progressive increase in activated caspase-3 (17 or 19 kDa) in the elimusertib -treated tumors, a marker of cell apoptosis and death. The activity of caspase-3/7 was significantly upregulated by elimusertib (0.3 μM) treatment for 24 hours (p<0.0001) and 48 hours (p<0.0001) as demonstrated by a luminescence-based caspase-3/7 activity assay (Figure 3.C).

Figure 3.

Representative western blot of CS cell lines treated with elimusertib. 3.A) ATR and p-ATR expression in nine primary CS cell lines. 3.B) SARARK-3 treatment with elimusertib significantly reduced the level of phosphorylated ATR and phosphorylated CHK1; levels of cleaved-caspase-3 increase with prolonged exposure demonstrating increased cell apoptosis. 3.C) Relative caspase-3/7 activity after 24 and 48 hours of elimusertib exposure vs. non-treated cells. The activity of caspase-3/7 was significantly upregulated after both 24 and 48 hours (p<0.0001). 3.D) ATR and p-ATR expression levels by HRD status. HRD cell lines express significantly higher levels of ATR (p=0.0379) and p-ATR (p=0.0049) proteins.

We next evaluated the expression levels of ATR and p-ATR of nine cell lines by western blot and HR status. When the HRD cell lines (SARARK3, SARARK 4, and SARARK9) are compared with the HRP cell lines (SARARK1, SARARK 7, and SARARK11, SARARK12, SARARK 13, and SARARK14), the HRD cell lines expressed significantly higher levels of ATR (p=0.0379) and p-ATR (p=0.0049) proteins (Figure 3.D). Taken together, these results demonstrate that elimusertib has the capability of inhibiting ATR activity and inducing apoptosis in CS cells, and that it may be more efficacious in HRD tumors (Figure 3).

3.4. Elimusertib shows strong antitumor activity as single agent in an HRD CS xenograft model

We next evaluated the impact of elimusertib in vivo in a SARARK 3 HRD CS xenograft model. As shown in Figure 4.A, mice exposed to elimusertib treatment exhibited a significantly slower rate of tumor growth compared to vehicle control starting at day 11 (p<0.0001). Furthermore, mice treated with elimusertib had significantly prolonged overall survival compared to mice treated with vehicle control. Median OS of mice treated with elimusertib was 39 days, while median OS of mice treated with vehicle control was only 11 days (Log Rank Mantel-Cox test p<0.0001) (Figure 4.B). Twice daily oral dose of elimusertib 20 mg/kg was well tolerated among mice, with no clear impact on body weight when compared to mice treated with vehicle control (Figure 4.C).

Figure 4.

Antitumor activity and overall survival in mice inoculated with HRD CS xenograft tumor model (PDX SARARK 3) after treatment with elimusertib (20mg/kg BID PO) compared to vehicle control. 4.A) Elimusertib demonstrates significant tumor growth inhibition compared to vehicle control (p<0.0001). 4.B) Overall survival was significantly prolonged among mice treated with elimusertib (p<0.0001). 4.C) Xenograft mice weight changes by treatment group did not differ.

4. Discussion

Uterine and ovarian CS are rare aggressive gynecologic tumors characterized by high recurrence rates after initial surgery, (i.e., 50-80%) limited response to standard chemotherapy regimens, and poor survival regardless of stage at presentation [1,3]. Recent work has demonstrated that many CS tumors have genetic signatures that confer some level of HRD and other DNA damage repair deficiencies [4,6]. In tumors with mutations in double strand break repair and DNA damage repair, the ATR pathway becomes more vital for cell survival. [6]. In this study, we demonstrate that the ATR pathway may represent a promising novel target for the treatment of patients with uterine and ovarian CS.

Cancer cells survive genomic instability by relying heavily on DNA damage repair pathways, one of which is the ATR pathway. ATR and ATM kinases are the two key upstream sensors of DNA damage [15,16]. ATM is activated in response to primarily double strand DNA breaks (DSB), while ATR responds to a larger swath of DNA damage, including DSB, single strand DNA breaks (SSB), and other replicative stress. ATM is also involved in the recruitment and activation of ATR.[17] The main downstream target of ATR is the checkpoint kinase 1 (CHK1), which is activated by phosphorylation and, in turn, inactivates (by phosphorylation) the phosphatases CDC25A and CDC25C, which are involved in the dephosphorylation and activation of cyclin dependent kinases 2 (CDK2) and 1 (CDK1), respectively [18] The maintenance of CDKs in their phosphorylated and inactivated form prevents S and M phase entry and consequently blocks cell cycle division, providing time for cells to undergo either accurate DNA repair or apoptosis if DNA damage is too extensive [19].

Prior preclinical and early clinical studies have demonstrated ATR inhibition to be efficacious in combination and as a single-agent therapy in a variety of human solid tumors and blood cancers that harbor mutations affecting the DNA damage repair pathway, and in those harboring HRD signatures [10,20,21]. In this regard, mutational signatures have recently been demonstrated to provide major insights into the biological processes shaping the tumor genome [22,23] and can potentially inform patients’ sensitivity or resistance to targeted treatments in multiple tumor types [24–26]. We utilized our recently published WES data from a cohort of fresh-frozen CS tumor samples which demonstrated the presence of HRD-signatures and derangements in multiple genes of the DNA damage repair pathway in a large subset of ovarian and uterine CS. When we analyzed genes involved in homologous recombination and DNA damage repair pathways among the primary CS cell lines used in this study (including but not limited to ATR/ATRX/DAXX/ATM/ARID1A), we found several to be altered (aberrations in SNV, CNV or both).

Interestingly, the cell line harboring gain of function (GOF) in ATR and DAXX (i.e., SARARK 13), was found to be more resistant to elimusertib in vitro, while tumors harboring loss of function (LOF) in ATRXand/or ARID1A genes (i.e., SARARK 1 and SARARK 11) demonstrated sensitivity to elimusertib, consistent with prior reports of alterations that confer sensitivity to ATRi [10,25]. Our findings in uterine and ovarian CS support the idea that evaluating the character of mutations in DDR genes (i.e., GOF/LOF) can help identify tumors that may be sensitive or resistant to ATRi. It could be hypothesized that amplifications (i.e., GOF) in DDR genes (i.e., ATM, ATR, DAXX) suggest resistance to ATRi, while LOF mutations in DDR pathways (i.e., ATR, ATRX, ARID1A) confer sensitivity to ATRi. These speculations are consistent with studies showing that GOF mutations in the DAXX pathway correlate with tumorigenesis, disease progression, and treatment resistance in a variety of human tumors [27–29]. This hypothesis is further supported by recent submitted data from our group in which patient-derived-xenograft (PDX) models of uterine leiomyosarcoma harboring ATRX SNV LOF led to both decreased expression of ATRX protein, and higher sensitivity to ATRi [30].

In the study described here, we also provide evidence that WES-extracted mutational signature-3 (HRD-related) in uterine and ovarian CS is correlated with preclinical sensitivity to elimusertib. Indeed, all available CS cell lines with a pure high-grade serous epithelial component (the driving force of CS [31]), that were HRD were found to be highly sensitive to ATR inhibition in vitro and in vivo. In contrast, primary CS cell lines with a pure high-grade serous epithelial component harboring a HRP signature were found resistant to elimusertib. While this subgroup analysis is limited by the relatively small number of primary CS having these histologic characteristics, our results suggest a significantly higher activity of elimusertib in HRD CS with a high-grade serous component, which account for the overwhelming majority of uterine and ovarian CS [4]. These results are also consistent with a recent report from our group demonstrating that primary OCS and UCS cell lines harboring HRD signature-3 are significantly more sensitive to PARPi (i.e., olaparib) in in vitro and in vivo experiments when compared to HRP CS [6]. Of interest, 4 HRP CS cell lines harboring pure or mixed endometrioid components demonstrated sensitivity to elimusertib in vitro similar to that detected in HRD cell lines. While the reason for these results is currently poorly understood, previous studies have demonstrated a differential sensitivity of cancer patients harboring diverse histologic subtypes of ovarian cancer and uterine cancers to chemotherapy [32,33].

Western blot analysis demonstrated a higher level of ATR expression in HRD compared to HRP CS. Moreover, when we investigated downstream targets affected by elimusertib in HRD CS cell lines, we were able to confirm a dose dependent inhibition of ATR, p-ATR and its downstream effector p-CHK1 starting 30 min after dosing and lasting for up to 48h after treatment. Importantly, cleaved caspase 3 levels, an apoptosis marker, showed a dose-dependent increase in expression at 24 and 48 hours after treatment. Finally, when we evaluated the activity of elimusertib in animals xenografted with a HRD cell line (i.e., SARARK3), our results confirmed those of the in vitro experiments. We demonstrate that administration of elimusertib led to significant impairment of SARARK3 tumor growth and significant increase in overall survival compared to controls. Additionally, elimusertib did not cause increased toxicity compared to vehicle controls. Due to the rarity of carcinosarcoma, our in vivo model is from only one ovarian CS cell line (SARARK3), which is HRD. While this represents a limitation of our study, the inclusion of 9 cell lines (a mix of UCS and OCS with varied HRD status) in our in vitro experiment is a strength and increases the generalizability of our findings. Taken together, these results demonstrate that elimusertib effectively inhibits ATR activity and induces apoptosis in CS models. Further evaluation of the possible link between this HRD mutational signature and sensitivity to ATR inhibitors may be warranted in clinical studies. In conclusion, we report for the first time the remarkable in vitro and in vivo activity of elimusertib against biologically aggressive CS models. Given that a significant subset of OCS and UCS may express HRD signatures [6] or harbor alterations in genes involved in the DDR pathway, (i.e., ATR/ATRX/DAXX/ATM/ARID1A), which are associated with higher sensitivity to ATRi, our data supports development of phase II trials of elimusertib for patients with advanced or recurrent CS resistant to chemotherapy.

Highlights.

• Elimusertib (BAY1895344), a novel ATR kinase inhibitor, demonstrates anti-proliferative and anti-tumor activity against uterine and ovarian carcinosarcoma

• Elimusertib showed dose-response cytotoxicity against carcinosarcoma cell lines in vitro

• Elimusertib decreased tumor growth in ovarian carcinosarcoma PDX mouse models with homologous recombination deficiency signatures

• Elimusertib prolonged overall survival in PDX mouse models of ovarian carcinosarcoma and did not add toxicity

• Tumor cells exposed to Elimusertib showed decreased phosphorylated-ATR and increased apoptotic molecules on western blot