Feasibility IB trial of paclitaxel/carboplatin + Galunisertib. (a small molecule inhibitor of the kinase domain of type 1 TGF-B receptor) in patients with newly diagnosed, persistent or recurrent carcinosarcoma of the uterus or ovary

Abstract

Objective.

Carcinosarcoma (CS) is a rare, aggressive malignancy characterized by high rates of extrauterine spread, frequent recurrences, and poor prognosis. Until recently, the standard of care treatment included surgery and chemotherapy given in an adjuvant or metastatic setting. This study assessed safety and tolerability of combining carboplatin(C) and paclitaxel(T) with the TGFb-R1 inhibitor galunisertib (GB).

Methods.

In this IRB approved (NCT03206177) feasibility study, eligible patients (pts) included those diagnosed with uterine or ovarian CS for whom treatment with CT is planned. A safety lead of CT plus GB 80 mg days 4–17 on a 28-day cycle, and if no dose-limiting toxicities (DLTs), the full dose of GB 150 mg BID would be used. The primary endpoint was completing 4 cycles of combination therapy without DLT.

Results.

3 patients were enrolled in the safety lead and 21 pts. were treated with CT and GB 150 mg po BID combination for a total of 24 patients. Among them, 81 % had stage III/IV and 54.2 % had measurable disease (MD). No DLTs were reported in the safety lead in. Among the 21 pts. treated at 150 mg po BID 1/21 had a DLT that compromised completion of 4 cycles. Therefore, the combination was deemed feasible. Among the 13 pts. with MD, 31 % had a partial response and 15 % had stable disease. The median progression-free survival (PFS) was 6.6 months (95 % CI: 5.4–10.8 months), and the median overall survival (OS) was 18.9 months (95 % CI: 10.81 -NR). 75 % had ≥ grade 3 treatment-emerged adverse events, the most common was neutropenia (41.7 %).

Conclusions.

GB with CT was well-tolerated with 1 DLT noted. The study was prematurely stopped when the development of GB was discontinued, but targeting TGFb downstream signaling remains of interest.

1. Introduction & background

Uterine carcinosarcoma (UCS) is a rare but highly aggressive malignancy characterized by biphasic histology, containing both epithelial and mesenchymal components [1]. UCS accounts for less than 5 % of all uterine cancers, yet it contributes disproportionately to uterine cancer-related mortality due to its high metastatic potential and resistance to standard therapies [2–4].

Despite advances in surgical techniques and systemic treatments, prognosis remains poor, with a five-year survival rate of less than 35 %, highlighting the need for novel therapeutic strategies [5]. Increasing evidence suggests that transforming growth factor-beta (TGF-β) signaling plays a pivotal role in UCS pathogenesis, making it an attractive therapeutic target [6–8].

TGF-β is a pleiotropic cytokine involved in cell proliferation, differentiation, and immune regulation. In UCS, TGF-β signaling is frequently upregulated, promoting epithelial-to-mesenchymal transition (EMT), tumor immune evasion, and metastatic dissemination [9]. Preclinical studies have demonstrated that elevated TGF-β activity enhances tumor cell plasticity and invasiveness, contributing to the aggressive phenotype of UCS [10]. Additionally, genomic analyses have identified recurrent alterations in key components of the TGF-β pathway, such as TGFBR2 and SMAD4, in UCS tumors, reinforcing its role as a critical driver of oncogenesis [11,12].

Given the central role of TGF-β in UCS pathogenesis [7,8], several therapeutic strategies have emerged to target this pathway, including TGF-β receptor kinase inhibitors, ligand traps, and SMAD-targeted therapies [13,14]. Preclinical studies have demonstrated that TGF-β inhibition can reverse EMT, enhance anti-tumor immunity, and improve sensitivity to existing treatments, including chemotherapy and immune checkpoint inhibitors [15,16]. Furthermore, combination approaches integrating TGF-β inhibitors with PD-1/PD-L1 blockade or cytotoxic agents have demonstrated promise in overcoming UCS resistance mechanisms [17,18].

Given the aggressive nature of UCS and its poor response to conventional therapies, targeting the TGF-β signaling pathway represents a promising therapeutic strategy. TGF-β signaling drives EMT, promoting loss of epithelial markers (E-cadherin) and upregulation of mesenchymal markers such as N-cadherin, vimentin, SNAIL, and TWIST in ovarian cancer [19,20] as well as contributing to tumor plasticity, invasion, and metastasis [9,21]. High TGF-β activity correlates with poor chemotherapy response in ovarian and other gynecologic CSs.

Galunisertib (GB) was the first small molecule inhibitor (SMI) of TGFβ to be clinically investigated in patients. The first-in-human phase I study identified a safe therapeutic window of up to 300 mg/day intermittent administration (14 days on and 14 days off on a 28-day cycle), allowing further clinical investigation [22]. Preclinical and first clinical results suggested a role for GB in patients with glioblastoma, hepatocellular carcinoma, and pancreatic cancer [13]. Further, a phase Ib study with gemcitabine found no dose limiting toxicity, and the 300 mg/day dose was advanced into a phase II study in patients with metastatic pancreatic cancer [23].

Preclinical work done in our lab, demonstrated that the components of the TGFβ pathway are expressed and functional in UCS tissue samples and cell lines [7]. Our preclinical work further clarifies the role of TGF-β signaling in CS, demonstrating that mRNA expression of TGFβ-I, TGFβ1-II, TGFβR-I, TGFβR-II, and c-MYC was significantly higher in tumors that recurred compared to those that did not. Additionally, TGF-β activation induced: phosphorylation of SMAD2/3 and JNK1, nuclear localization of NFAT1, c-MYC transcriptional activation.

These findings suggest that TGF-β signaling plays a crucial role in tumor recurrence and progression in CS. Furthermore, LY2157299 GB efficiently blocked TGF-β-induced SMAD2/3 activation, proliferation, migration, and EMT, supporting its potential therapeutic role in CS [7,8]. Based on the association of TGF-β signaling with EMT in UCS [7], we proposed a phase Ib feasibility trial with GB in combination with CT in patients with OCS or UCS.

2. Methods

This is an IRB-approved clinical trial (NCT03206177). The objectives of this study were to determine the feasibility (safety and tolerability) of combined CT plus GB in patients with UCS or ovarian carcinosarcoma (OCS). Additionally, this study sought to determine the frequency and severity of AE assessed by CTCAE v.4. Toxicities were attributed to GB versus chemotherapy based on investigator assessment using standardized criteria (CTCAE v.4), the timing of adverse event onset relative to treatment phase, and the known toxicity profiles of each agent. The study was designed to have a safety lead-in of CP plus GB at a reduced dose of 80 mg po BID for up to 4 patients, which was then escalated to full dose GB at 150 mg po BID, as will be described below.

Eligible patients were those with a diagnosis of UCS or OCS for whom treatment with carboplatin and paclitaxel was planned. Surgery was allowed as part of the treatment, but was not mandatory for inclusion. Surgery was allowed as part of the treatment, but was not mandatory for inclusion. Patients had to have ECOG performance status 0–2 and adequate bone marrow, renal, and hepatic function. Both measurable and non-measurable diseases were allowed, and all FIGO stages were eligible.

2.1. Chemotherapy

Patients post safety lead in were treated with T 175 mg/m2 over 3 h, IV days 1, C AUC 6 over 1 h, IV days 1, and GB 150 mg po BID day 4–17. The therapy cycles were 28 days in length. In patients with a history of prior pelvic radiation (to reduce the risk of severe toxicity), the initial dose of C was AUC 5, and the T dose was 135 mg/m2. Patients received combination therapy until unacceptable toxicity or progression. At any point beyond completion of the first 4 cycles of combination therapy, if the treating physician 102 believes cytotoxic chemotherapy may be safely suspended, patients could continue GB 150 mg po BID as maintenance therapy until progression or unacceptable toxicity.

For the safety lead in, patients were treated with T 175 mg/m2 over 3 h, IV days 1, C AUC 6 over 1 h, IV days 1, and GB 80 mg po BID day 4–17. For these 3 patients, dose limiting toxicities were assessed through cycle 4 and if none were reported, the feasibility study launched at GB 150 mg po BID as above.

2.2. Statistical considerations

2.2.1. Sample size calculation

Given that this is a feasibility study, the rate of completing 4 cycles of combined CT and GB without DLT is considered the primary outcome. With 25 evaluable patients (i.e., excluding patients who will terminate the treatment for reasons other than toxicity) and an expected completion rate of 60 %, the lower bound of the two-sided exact Clopper-Pearson 90 % confidence interval (CI) would be 42 %, which provides sufficient evidence that the combined regimen is feasible for inclusion in a randomized phase II study. The 60 % rate of completion of 4 cycles was gleaned from historical UCS data sets. Using more recent studies, the expected rate of completion of 6 cycles is approximately 85 % [24–28].

DLT-qualifying events included: Dose delay of greater than 3 weeks due to failure to recover counts, febrile neutropenia, grade 4 neutropenia lasting >7 days, grade 4 thrombocytopenia or bleeding associated with grade 3 thrombocytopenia, study treatment-related grade 3 or 4 non-hematological toxicity (excluding alopecia, fatigue, hypersensitivity reaction, nausea, vomiting, constipation, diarrhea, hypokalemia, hypomagnesemia, hypocalcemia, hypophosphatemia) or any drug-related death.

2.2.2. Statistical methods

Data analysis was largely descriptive and exploratory. The Clopper-Pearson method reported the completion rate with an exact 90 % CI (2-sided). The frequency and severity of AE are tabulated. Median PFS and OS (and CIs) for patients who receive the combined regimen were computed using the Kaplan-Meier method. All analyses were performed using SAS software (version 9.4, Cary NC).

2.3. Translational analysis (Immunohistochemistry)

For patients with tissue available, immunohistochemistry for pSMAD2, cMYC, Vimentin, and Cytokeratin was conducted at the Stephenson Cancer Tissue Pathology Core using a Leica multi-stainer (ST5020) following standard protocols. The antibodies used following cell block creation included pSMAD2 (Cell Signal Technology, S3108, 1:200), c-MYC (Abcam, ab32072, 1:200), Cytokeratin AE1/AE3 (Abcam, ab27988, 1:20), and Vimentin (Cell Signal Technology, S5741, 1:300). Pathologists evaluated slides for pSMAD2 and c-MYC expression. H&E, Cytokeratin AE1/AE3, and Vimentin staining were used to assess epithelial and mesenchymal components.

3. Results

This clinical trial anticipated an accrual of 36 subjects with 25 evaluable patients for the primary feasibility endpoint. However, the study was terminated prematurely due to the sponsor discontinuation of GB development. This study reports on the 24 patients who were enrolled.

At the time of study closure, 24 patients were enrolled and treated on this trial. Of these 24, 3 were treated in the safety lead in and 21 were treated at the 150 mg dose and evaluable for feasibility assessment. One patient was non-evaluable for the feasibility endpoint, due to taking insufficient drug but is included for efficacy and safety analysis. Demographics and tumor characteristics for the entire cohort are outlined in Table 1. Noteworthy characteristics of this cohort are that for 87 % of participants, this trial was primary therapy for their cancer, 24 % of the participants were non-White, and almost 90 % were treated for UCS.

Table 1.

Demographic data.

	Statistic	Total (N = 24)
Age	n	24
	Mean(SD)	69.54 (9.833)
	Median	71.5
	IQR (25th,75th percentile)	59,71
	Min, Max	41,82
BMI	n	23
	Mean(SD)	27.08 (8.401)
	Median	28.0
	IQR (25th,75th percentile)	26.5, 38.7
	Min, Max	1.45, 36.91
Ethnicity
Hispanic or Latino		1 (4.2)
NOT Hispanic or Latinos		23 (95.8)
Race
American Indian/Alaska Native		1 (4.2)
Asian		2 (8.3)
Black or African American		3 (12.5)
Unknown/ not reported		1 (4.2)
White		17 (70.8)
Prior chemotherapy
No		21 (87.5)
Yes		3 (12.5)
Prior radiation
No		22 (91.7)
Yes		2 (8.3)
Prior surgery
No		22 (91.7)
Yes		2 (8.3)
Primary site
Ovary		3 (12.5)
Uterine		21 (87.5)
FIGO at diagnosis
IA		4 (16.7)
IB		1 (4.2)
III		1 (4.2)
IIIC		8 (33.3)
IV		9 (37.5)
Missing		1 (4.2)
Baseline ECOG performance status
0		11 (45.8)
1		12 (50.0)
Missing		1 (4.2)
Mismatch repair status (n = 21 uterine)
Proficient		20
Deficient		1
Unknown		0

In the safety lead in at 80 mg po BID of GB there were 3 patients treated and no DLTs noted in the first 4 cycles of combination therapy and no signal of any additive toxicity. For this reason – the decision was made to move to full dose GB evaluation rather than exposing another patient to a reduced dose.

In the expansion cohort of 21 patients at 150 mg po BID GB combination therapy, there was one patient as mentioned above who did not take at least 75 % of drug not related to adverse events and so was deemed not evaluable for the primary endpoint but is included in the overall safety and efficacy evaluation. There was 1 DLT qualifying event among the 20 evaluable patients and the remainder of the patients successfully completed 4 cycles of combination therapy without DLT qualifying events or other signals of additive toxicity. Therefore, the feasibility rate among patients who received 150 mg po BID GB combination therapy is 95 % (90 % Clopper-Pearson CI 78 % - 100 %). The DLT qualifying event occurred in a patient treated on the 150 mg BID dose cohort who developed pulmonary fibrosis grade 2; eleven months later, pulmonary infection, and one month after the infection, pulmonary fibrosis grade 5, for which the treating physician, GB, was possibly related.

Apart from completion of 4 cycles, the combination of GB with CT appeared well tolerated with only 8/24 (33 %) patients requiring dose delay (2 at the 80 mg dose and 6 at the 150 mg po BID dose of GB). 46 % of patients (11/24) experienced an adverse event determined to be serious (SAE) but none of these were deemed related to GB except for one patient with pulmonary fibrosis (in the setting of a pulmonary infection) which was deemed possibly related. Dose modifications were required in 4 % of cycles (all in one patient). Two modifications were for paclitaxel and one for carboplatin and all not related to GB. There were no treatment discontinuations due to toxicity.

Treatment emergent adverse events (TEAEs) of all grades and > grade 3 that occurred in >10 % of patients are summarized in Table 2. The most common TEAE was neuropathy (54 %, all grade 1 or 2), followed by fatigue (50 %, all grade 1 or 2) and neutropenia (41.7 % all grade 3 or 4).

Table 2.

Listing of treatment emergent adverse events occurring in ≥10 % of participants.

AE Term	Number of Patients who Experienced AE (N = 24)
	Overall		Grade 3+		Related to Galunisertib
	N	%	N	%	N	%
NEUROPATHY	13	54.2	0	0.0	0	0.0
FATIGUE	12	50.0	0	0.0	6	25.0
NEUTROPENIA	10	41.7	10	41.7	6	25.0
NAUSEA	9	37.5	1	4.2	3	12.5
ARTHRALGIA	8	33.3	0	0.0	0	0.0
UTI	7	29.2	1	4.2	0	0.0
RASH	5	20.8	0	0.0	2	8.3
ANEMIA	4	16.7	4	16.7	2	8.3
ANOREXIA	4	16.7	0	0.0	1	4.2
CONSTIPATION	4	16.7	0	0.0	1	4.2
THROMBOCYTOPENIA	4	16.7	3	12.5	2	8.3
BACK PAIN	3	12.5	0	0.0	0	0.0
DIZZINESS	3	12.5	0	0.0	2	8.3
DYSGEUSIA	3	12.5	0	0.0	0	0.0
DYSPNEA	3	12.5	0	0.0	0	0.0
FALL	3	12.5	1	4.2	0	0.0
HEADACHE	3	12.5	0	0.0	0	0.0
HYPOMAGNESEMIA	3	12.5	0	0.0	0	0.0
INTERMITTENT NAUSEA	3	12.5	0	0.0	1	4.2
VAGINAL BLEEDING	3	12.5	0	0.0	0	0.0
VOMITING	3	12.5	1	4.2	1	4.2

Efficacy endpoints were secondary endpoints and are summarized here. Among 13 patients with 155 measurable disease at study enrollment, the best response was partial response in 31 % (95 % CI 9 %–61.4 %). An additional 15 % of patients had stable disease, and the rest were with disease progression.

The median PFS was 6.6 months (95 % CI: 5.4–10.8 months), and the median OS was 18.9 months (95 % CI: 10.81 -NR). The time spent on study is displayed in the swim lane plot in Fig. 1.

Fig. 1.

Duration of treatment and response to paclitaxel, carboplatin and galunisertib by microsatellite status in uterine or ovarian carcinosarcoma. Each horizontal bar represents one patient, showing time on treatment from initiation (x-axis, in weeks). Bars are color-coded by microsatellite status: high microsatellite instability (MSI-H, red), microsatellite stable (MSS, blue), or unknown (gray). Symbols indicate key events: filled circles denote first radiologic assessment showing response, diamonds indicate disease progression, crosses mark death, and arrows represent patients with ongoing treatment at data cutoff.

3.1. Correlative endpoints

Six patients had paired biopsies for exploratory evaluation in this trial. Of interest were the assessments of changes in phosphorylated SMAD and changes in the epithelial component. These findings are outlined in the Table 3. With small numbers, interpretation of this data is limited however it is interesting that patient 3, who experienced the shortest overall survival of this cohort, displayed pre-treatment tumor that was 30 % epithelial and 70 % mesenchymal, 25 % of the epithelial cells and 10 % of mesenchymal cells were pSMAD2 positive and 0 % of the epithelial cells and 90 % of the mesenchymal cells were cMYC positive. After therapy, there was a reduction in the epithelial component (from 30 % to 5 %), an increase in the mesenchymal component (from 70 % to 95 %), 0 % of the epithelial cells and 90 % of the mesenchymal cells were pSMAD2 positive and there was a substantial decrease in cMYC levels in the mesenchymal component (from 90 % to 0 %). The high mesenchymal component at the beginning (pre-treatment) became even higher post-therapy. 90 % of mesenchymal cells were SMAD positive, inducing tumor-promoting effects, while c-MYC (one of the most significant factors for expressing tumor suppressor effects) was prominently decreasing.

Table 3.

Summary of changes in epithelial and mesenchymal proportion as well as cMYC and pSMAD2 immunohistochemistry in pre and post galunisertib biopsies.

Sample ID	Treatment group	%epithelial	%mesen chymal	cMYC		pSMAD2		Cycles	Response to therapy
				Epit pos%	Mes pos%	Epit pos%	Mes pos%
PT 1	Pre-treatment	95	5	80	0	80	0
	Post-treatment	60	40	0	0	0	0	6	Partial response
PT 2	Pre-treatment	50	50	0	0	50	50
	Post-treatment	10	90	90	<10	95	95	2	Disease progression
PT 3	Pre-treatment	30	70	0	90	25	10
	Post-treatment	5	95	<1	0	0	90	2	Disease progression
PT 4	Pre-treatment	20	80	10	0	80	60
	Post-treatment	80	20	0	0	90	20	10	Stable disease
PT 5	Pre-treatment	80	20	0	0	60	50
	Post-treatment	50	50	0	0	90	70	14	Not evaluable
PT 6	Pre-treatment	95	5	0	0	95	0
	Post-treatment	99	1	0	0	80	0	7	Partial response

4. Summary and discussion

This study demonstrated the feasibility and safety of combining GB, a TGF-β receptor I inhibitor (TGFβ-RI), with CT in CS. Importantly, there was no indication that the AE profile was negatively impacted by the combination. Additionally, exploratory correlative analyses provided evidence of TGFβ-RI inhibition on EMT and disease progression, rein forcing the role of TGFβ signaling in CS pathogenesis.

4.1. TGF-β and EMT in carcinosarcoma

EMT plays a crucial role in cancer progression, stemness, immune evasion, and metastasis [10,29]. By inducing EMT, tumor cells acquire invasive potential, allowing for detachment from the basement membrane, tissue infiltration, and metastatic dissemination. EMT is also a well-established mechanism of drug resistance, particularly in aggressive malignancies such as CS. Loss of E-cadherin, a key epithelial adhesion molecule, is a hallmark of EMT and contributes to tumor cell detachment and metastatic spread [30].

Although the role of EMT in CS has been relatively understudied, Saegusa et al. demonstrated a complete shutdown of E-cadherin expression in the mesenchymal component of CS, as well as increased pAKT signaling and nuclear β-catenin levels, indicating activation of pathways associated with EMT and tumor progression [31]. Similarly, Romero-Perez et al. identified differential expression of EMT-related genes in CS compared to endometrial carcinoma, including overexpression of HMGA2, a gene that cooperates with the TGF-β/SMAD pathway to drive EMT [32].

TGF-β is a multifunctional cytokine that regulates EMT and has dual roles in cancer: it suppresses tumor growth in early-stage disease while promoting tumor progression, invasion, and metastasis in advanced malignancies [33]. Through both SMAD-dependent and non-SMAD pathways, TGF-β upregulates EMT-associated transcription factors (SNAIL, SLUG, TWIST, and ZEB1/2), driving the transition to a more aggressive, mesenchymal phenotype [34].

Additionally, TGF-β signaling has been implicated in chemoresistance. In triple-negative breast cancer (TNBC) models, taxol treatment induced autocrine TGF-β signaling, and inhibition of TGF-β prevented tumor recurrence, suggesting that combining TGF-β inhibitors with chemotherapy could enhance treatment efficacy [35]. Similarly, in ovarian cancer, TGF-β1 expression was associated with chemotherapy sensitivity and prognosis, further supporting its role as a potential biomarker and therapeutic target [36].

TGF-β signaling promotes tumor invasion and immune evasion by inducing an immunosuppressive tumor microenvironment. It enhances the activity of regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs), while inhibiting the activity of cytotoxic T cells and natural killer (NK) cells, thereby reducing anti-tumor immunity. Given these immunosuppressive effects, TGF-β blockade may enhance the efficacy of immune checkpoint inhibitors (CPIs) [37].

4.2. Future directions and challenges

While GB development has been discontinued, several TGF-β-targeting agents remain under investigation. The approval of CT + Immune checkpoint inhibitor (ICI) for metastatic CS represents an important advancement, though responses in pMMR tumors remain modest, highlighting the need for additional therapeutic strategies [38–40]. Approximately 10–20 % of UCS exhibit deficient mismatch repair (dMMR) or microsatellite instability high (MSI-H) status. Studies analyzing UCS tumor samples have shown that MLH1 promoter hypermethylation is the most common mechanism of dMMR in these tumors, similar to endometrial carcinomas [41,42].

Given the role of ICI (e.g., pembrolizumab, dostarlimab) in treating dMMR/MSI-H tumors, this subset of UCS may benefit from immune-based therapies. However, the majority (~80–90 %) of UCS are mismatch repair proficient (pMMR) and may require alternative therapeutic strategies.

Given the immunosuppressive role of TGF-β, its inhibition may enhance response to CPIs, particularly in tumors with an inherently poor response [37].

Preclinical studies in pancreatic cancer models demonstrated improved survival and increased T-cell infiltration with GB + ICI therapy, but clinical results were less promising. Melisi et al. (NCT02734160), evaluating durvalumab + GB, the disease control rate was only 25 %, with a median OS of 5.72 months and a PFS of 1.87 months [43]. These results suggest that while TGF-β inhibition is mechanistically compelling, optimal therapeutic combinations remain uncertain.

Several alternative strategies have been explored:

1. TGF-β Inhibitors (e.g., vactosertib) – These tyrosine kinase inhibitors selectively target TGF-βRI, though toxicity concerns limit their ability to inhibit all TGF-β isoforms.

2. Bifunctional Fusion Proteins (e.g., bintrafusp alfa) – These agents target TGFβR-II and PD-L1 simultaneously, but clinical efficacy has been low to modest, with biomarker-driven patient selection needed for improved outcomes [44].

5. Conclusion

Despite its critical role in CS progression, no TGF-β inhibitors have been FDA-approved for cancer therapy. Future clinical development of TGF-β blockade will likely require combination strategies, integrating immune checkpoint inhibitors, chemotherapy, or targeted therapies. The knowledge accumulated over the past 15 years of TGF-β pathway research will be essential in designing more effective and safer treatments.

To our knowledge, this was the first clinical trial evaluating a TGF-β inhibitor in endometrial cancer. Given its biphasic histology and strong EMT driven biology, uterine carcinosarcoma (UCS) represents an ideal model for studying EMT and assessing therapeutic strategies targeting this pathway.

The TGFβR-I inhibitor (GB) was well tolerated in combination with paclitaxel and carboplatin. However, the limited availability of paired biopsies precluded definitive conclusions regarding biomarker responses. Despite this, the mechanism of action of GB and its tolerability support further investigation of TGF-β inhibition, particularly in combination with immune checkpoint inhibitors (ICIs) in preclinical models.

This study has several limitations, including a small sample size, limited paired biopsies, and the use of a therapeutic agent that was later discontinued due to limited efficacy across solid tumors. Additionally, the requirement for GB dosing every 28 days necessitated a non-standard paclitaxel/carboplatin schedule, though this likely had minimal impact on efficacy. Moving forward, next-generation TGF-β inhibitors should be evaluated within standard chemotherapy dosing regimens and potentially ICI to optimize clinical applicability.

HIGHLIGHTS.

• Carcinosarcoma (CS) of the uterus or ovary is an aggressive malignancy with a poor prognosis.

• CS histogenesis and phenotypical changes have been explained with epithelial-mesenchymal transition (EMT).

• Transforming growth factor beta (TFG-β) plays an essential role as an inducer of EMT and could be used as a targeted therapy.

• Galunisertib, combined with paclitaxel and carboplatin, was well-tolerated in patients with uterine and ovarian CS.

Data availability

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