Efficacy of cyclin-dependent kinase 4/6 inhibitors in preclinical adult-type ovarian granulosa cell tumor models

Abstract

Background.

Cyclin-dependent kinase 4/6 inhibitors (CDK4/6i) improve outcomes in metastatic, hormone receptor-positive breast cancer and low-grade serous ovarian carcinoma. Adult granulosa cell tumors (AGCTs) are similarly hormonally active tumors. We aimed to determine whether CDK4/6i were effective in pre-clinical models of AGCTs.

Methods.

Pre-clinical models used in this study include the KGN cell line, in cell culture and mouse models, and multiple novel AGCT tumoroids treated with CDK4/6i, including abemaciclib. Half-maximal inhibitory concentration (IC50) was used to assess potency of drugs. Cell number was determined using CCK8 cytotoxicity assay. Mice were treated with daily gavage with control or abemaciclib. Patients with AGCTs and prior CDK4/6i treatment were identified retrospectively.

Results.

CDK4/6i demonstrated high potency in KGN cells with an IC50 value of 50 nM for abemaciclib. Adding fulvestrant to abemaciclib did not significantly affect cell proliferation. Abemaciclib demonstrated proliferation inhibition in seven AGCT tumoroids. The xenograft model demonstrated significant differences in tumor volume (p = 0.003) and tumor mass (p = 0.0002) after treatment with abemaciclib or vehicle control. Eleven patients with AGCT who received CDK4/6i treatment were identified. Best responses included partial response (3, 27.3 %), stable disease (6, 54.5 %), and progressive disease (2, 18.2 %). The median duration of response was 6 months. The regimen was well tolerated.

Conclusions.

CDK4/6i demonstrated potent inhibition of tumor cell proliferation in multiple preclinical models. CDK4/6i demonstrated tolerability and effectiveness in a small cohort of patients with AGCT. This data supports the design of clinical trials to test the efficacy of CDK4/6i in the treatment of recurrent AGCT.

1. Introduction

Granulosa cell tumors (GCTs) are uncommon tumors that account for 2–5 % of ovarian cancers [1]. Most of these are adult-type GCTs (AGCTs), comprising 95 % of GCTs. AGCTs are driven by a pathognomonic missense mutation in FOXL2 (c.C402G; p.C134W) [2]. Compared to the more common epithelial ovarian tumors, AGCTs have improved outcomes. Most AGCTs are effectively treated with surgery alone at diagnosis, but approximately 30 % of cases will recur [3]. The preferred treatment for recurrent AGCTs is cytoreductive surgery [4] with or without the addition of either systemic chemotherapy, endocrine therapy, or radiation [5–11]. Emerging evidence also suggests the potential efficacy of immunotherapy in treating recurrent AGCTs [12,13]. Despite these advancements, the rarity of AGCTs limits the availability of randomized controlled trials, leaving the optimal treatment strategies undefined. Moreover, clinical molecular profiling has not identified a high prevalence of actionable targets in these tumors, underscoring the need for improved therapeutic approaches [14,15].

AGCTs are hormonally active and express high levels of estrogen (ER) and progesterone (PR) receptors [16]. As a result, anti-estrogen therapies, particularly aromatase inhibitors, have been a cornerstone in treating recurrent AGCTs. Retrospective data and expert opinion support their use [10], but a clinical trial has shown limited efficacy. The PARAGON/ANZOG-0903 trial evaluated the use of single-agent aromatase inhibitors in recurrent or metastatic AGCTs. Only 2.6 %, 1 out of 38 evaluable patients, had an objective response to treatment at 12 weeks [9]. This highlights the urgent need for more effective treatments for recurrent and metastatic AGCTs.

Cyclin-dependent kinase 4/6 (CDK4/6) inhibitors have become an essential component of treatment in many hormone-responsive malignancies. Translational research has shown that exogenous estrogen promotes transcription of cyclin D1–3, leading to CDK4/6-dependent phosphorylation of RB1 and promotion of the cell cycle [17]. Recent developments in the treatment of hormone receptor-positive, HER2-negative breast cancer with CDK4/6 inhibitors and anti-estrogen therapies have shown significant improvements in progression-free survival (PFS) and overall survival (OS). These improvements have been noted in the neoadjuvant [18,19], adjuvant [20,21], and recurrent or widely metastatic setting [22–25]. Similarly, low-grade serous ovarian cancer, another hormone receptor-positive malignancy [26], has shown promise with CDK4/6 inhibitors in multiple settings [27,28]. CDK4/6 inhibitors are often used in combination with anti-estrogen therapies. Fulvestrant, a selective estrogen receptor degrader, is effective with CDK4/6 inhibitors in both breast and low-grade serous ovarian cancer [27,29]. There are currently 3 FDA-approved CDK4/6 inhibitors for the treatment of breast cancer: abemaciclib, palbociclib, and ribociclib [30].

Given the hormone receptor-positive nature of AGCTs, CDK4/6 inhibitors are a compelling potential therapeutic. Our lab has developed novel model systems that generate high-quality preclinical data using tumoroid models from patient AGCT samples [31,32]. This study aimed to determine whether CDK4/6 inhibitors, both as monotherapy and in combination with fulvestrant, were effective in high-fidelity preclinical models of AGCTs.

2. Materials and methods

2.1. Cell lines

The KGN cell line (RIKEN Bioresource Research Center, Cat# RCB1154), derived from a recurrent AGCT [33], was provided by the RIKEN Bioresource Research Center (Ibaraki, Japan) through the National BioResource Project of the Ministry of Education, Culture, Sports, Science, and Technology/Agency for Medical Research and Development under a material transfer agreement. KGN cells were cultured using previously described methods [32]. All cell lines were independently authenticated using short tandem repeat profiling and regularly screened for mycoplasma infection.

2.2. Human subjects

Tumor tissue and blood of patients diagnosed with AGCT undergoing surgery at M.D. Anderson Cancer Center were obtained by the Gynecologic Oncology Multidisciplinary Tumor Bank under an Institutional Review Board (IRB) approved collection protocol (#LAB02–188). Each patient provided written, informed consent for biospecimen collection and use for research applications. The specific use of these tumor bank samples for tumoroid culture and molecular analysis in this study was separately IRB-approved (#2021–0933). This protocol also approved the collection and use of protected health information for these patients.

Patients who had received a CDK4/6 inhibitor, either as monotherapy or dual therapy, as part of their AGCT treatment at our institution were retrospectively identified. All patients were enrolled in an institutional review board (IRB)–approved Rare Gynecologic Malignancy Registry (protocol #PA17–0586). Data regarding patients’ prior treatment and response to CDK4/6 inhibitors was collected.

2.3. Tumoroid cultures

A previously described AGCT tumoroid biobank was used for in vitro experiments [32]. All tumoroids were characterized by immunohistochemistry (IHC) for AGCT markers and were submitted to whole exome sequencing. Briefly, fresh AGCT tissue was obtained from the Gynecologic Oncology Multidisciplinary Tumor Bank on the day of surgery and placed into a warmed RPMI growth medium. Tumor cell suspensions were prepared using a modified version of a previously published method [31] for preparing suspensions from epithelial ovarian tumor material [32]. Tumoroid growth medium contained AdDF+++ (DMEM/F12 + 1 % L-Glutamax, + 1 % Penicillin/Streptavidin, + 15 mM HEPES pH 7.4) with Rock inhibitor (Selleckchem, Cat# S1049) + 2 % 50xB27 supplement (GIBCO, Cat# 17504–044), + Beta Estradiol (100 nM, Millipore-Sigma, Cat# E2257), + Activin A (10 ng/ml, Peprotech, Cat# 120–14P), + Wnt3a (10 ng/ml, Peprotech, Cat# 315–20).

2.4. AGCT tumoroid culture cell viability assays

To perform cell viability assays following drug treatment in tumoroid cultures, 96-well plates were precoated with a 1:1 mixture of Matrigel and tumoroid growth medium as described above. AGCT tumor cells were resuspended in the same medium, mixed at 1:1 ratio with Matrigel and the cell/Matrigel mixture was overlaid onto the precoated wells. 100 μl medium containing the experimental treatments or vehicle control were added and cells were cultured for 72 h. CCK8 substrate was added for the last 4 h of culturing. Absorbance at 450 nm was measured in a POLARstar Omega microplate reader. Drug effects were evaluated compared to vehicle DMSO (0.1 %) controls. Treatment with high-dose cytotoxic Puromycin (5 μg/ml, Invivogen) was used to obtain negative control samples for background subtraction in absorbance measurements.

2.5. Reagents

A high-throughput drug sensitivity screen was previously performed using a compound library of 2326 compounds (TargetMol, Cat# TML10_Approved drug library and Cat# 268 TML12_Epigenetics drug library). The Combinatorial Drug Discovery Program at Texas A&M University maintains this compound library and performs all drug screens. Welte et al. included the results of three CDK4/6 inhibitors in their drug screen [32]. A CDK4/6 inhibitor abemaciclib and fulvestrant were obtained from Selleckchem (Texas, USA). Both drugs were solubilized in dimethyl sulfoxide (DMSO) for in vitro use. Half-maximal inhibitory concentration (IC50) and growth rate inhibition at 50 % (GR50) were used to assess the potency of the drugs. GR50 represents the concentration of a drug that reduces the cell growth rate by 50 % compared to a control group, whereas IC50 represents the concentration of a drug that inhibits cell viability by 50 % compared to a control group. Specifically, the Hafner growth rate inhibition technique was used, which is a method of quantifying drug response in cell cultures, accounting for the variation in cell division rates [34]. For in vivo gavage, abemaciclib was diluted in 1 % hydroxyethylcellulose (HEC) in 25 mM phosphate buffer pH 2.0.

2.6. Xenograft mouse model

KGN tumor cells were subcutaneously injected in the flanks of 8–10 week-old female severe combined immunodeficiency (SCID) mice (NOD.Cg-Prkdc^scid /J; The Jackson Laboratory, Stock #001303) at a concentration of 8 × 10⁶ cells in 100 μl DMEM +100 μl Matrigel per mouse. Before injection, the injection site was prepared by topical application of hair-removal cream (Pharmapacks) and cleansing with 70 % ethanol. The animals were anesthetized using Isoflurane. Tumor cells were pre-mixed with Matrigel for 15 min in a syringe and then injected using a 25 1/2G needle. Tumor growth was monitored by measuring tumor width (w) and length (L) with calipers at specific time points, and tumor volume (v) was calculated using the formula: v = w² × L × 3.14/6. The investigators were blinded to the groups during measurements. Two investigators measured each tumor and the average was used to calculate the final tumor volume. Drug treatment with either 50 mg/kg abemaciclib in HEC or HEC alone started on day 12 after tumor injection and continued until day 33. After completion of treatment, animals were euthanized, and tumors were extracted and weighed. Unstained slides of the tumors were prepared from formalin fixed, paraffin embedded tissue blocks, and stained for SF1 (Invitrogen, clone N1665, dilution 1:3000). Animal health and safety were monitored according to the IACUC-approved protocol, including observing animals for signs of pain and distress throughout the study. No adverse events were reported.

2.7. Statistical analyses

For the xenograft model, with an expected difference in tumor size of 20 % between the CDK4/6 inhibitor and placebo groups, with an assumed standard deviation of 25 %, and an alpha level of 0.05, a power calculation indicated that a sample size of 11 mice in the treated group and 10 mice in the placebo group would be sufficient to detect a statistically significant difference with 80 % power. Therefore, 12 mice were allocated to each group. Statistical significance and p-values (n.s. p value >0.05; * p value ≤0.05; ** p value ≤0.01) for experiments were determined using t-tests and Mann-Whitney tests using GraphPad Prism version 10.0.0 for Windows, GraphPad Software, Boston, Massachusetts USA, www.graphpad.com. All statistical comparisons were two-sided, where applicable, and a p-value <0.05 was considered statistically significant.

3. Results

The high-throughput drug screening done by Welte et al. included three FDA-approved CDK4/6 inhibitors, abemaciclib, palbociclib, and ribociclib in KGN cell lines [32]. Fig. 1 demonstrates the Hafner growth rate index (GRI) [34] of KGN treated with CDK4/6 inhibitors measured on day 3 of incubation with the associated IC50 and growth-rate inhibition concentration (GR50). Similar IC50 and GR50 rates were seen between palbociclib and abemaciclib; abemaciclib was chosen for further experimentation.

Fig. 1.

CDK4/6 inhibitors antagonize the growth of AGCT. A, Dose-response curve for three FDA-approved CDK4/6 inhibitors in KGN cell lines at day 3 of incubation. Proliferation determined by DAPI count growth assays. B, IC50 compared to GR50 of three FDA-approved CDK4/6 inhibitors in KGN cell lines at day 3 of incubation. GRI = Hafner growth rate index [34].

KGN cell lines were next treated with a combination of abemaciclib with or without fulvestrant (Fig. 2). KGN treated with abemaciclib alone had an IC50 of 0.2 μM (Fig. 2B). When KGN was treated with both fulvestrant and abemaciclib in an estrogen-rich medium, increasing the dose of abemaciclib demonstrated significant differences in cell proliferation (Fig. 2C), but the addition of fulvestrant made no difference in cell proliferation (Fig. 2C).

Fig. 2.

Abemaciclib inhibits the growth of AGCT KGN cell line. A, Dose-Response curve for abemaciclib and fulvestrant treatment of KGN AGCT cell line. Error bars, SD of the mean. IC50, Half-maximal inhibitory concentration B, Dose-response curve for abemaciclib treatment of KGN AGCT cell line. Error bars, SD of the mean. IC50, Half-maximal inhibitory concentration. C, CCK-8 quantification of abemaciclib, fulvestrant, and DMSO (control) treated KGN AGCT cell line. Results of a two-sided t-test are shown. * p value ≤0.05; ** p value ≤0.01.

Seven tumoroid models were used to determine the efficacy of abemaciclib with or without fulvestrant (Fig. 3). The clinical features of the donor patients are shown in Table 1. The majority of tumoroid samples came from patients with recurrent AGCT. There are a range of clinical features, including stages I-III, site of tumor collection, and different lines of chemotherapy, targeted therapy, and endocrine therapy. FOXL2 mutations were present in all specimens. GCT025 and GCT032 had TERT promoter mutations, and GCT029 had a TP53 mutation. Of the four patients with somatic tumor testing, none had mutations associated with CDK4/6 inhibitor resistance. Next-generation sequencing (NGS) results of tumoroid models can be found in supplemental table S6 of Welte et al. [32].

Fig. 3.

Bright field images of AGCT tumoroid cultures.

Table 1.

Clinical information of patients with AGCT tissue used in this study.

Tumoroid Number	Type	Years since diagnosis	Prior lines of systemic treatment	Types of prior systemic treatment	Initial Stage	Anatomic Site
GCT025	Recurrent	16	1	Chemotherapy	3A1	Pelvic peritoneum
GCT027	Primary	-	0	None	1 A	Ovary
				Chemotherapy
GCT028	Recurrent	15	4	Endocrine therapy	1C1	Diaphragm
				Targeted therapy
GCT029	Recurrent	11	0	None	2B	Rectosigmoid
GCT030	Recurrent	12	1	Endocrine therapy	1C1	Spleen
				Chemotherapy
GCT031	Recurrent	22	5	Endocrine therapy	2C	Pelvic peritoneum
				Targeted therapy
GCT032	Recurrent	19	4	Chemotherapy	1 A	Pelvic peritoneum
				Endocrine therapy

Two tumoroid models were treated with fulvestrant and abemaciclib. While abemaciclib showed a significant decrease in AGCT tumoroid viability, fulvestrant did not significantly decrease AGCT viability. Therefore, the remaining 5 tumoroid models were treated with abemaciclib alone, shown in Fig. 4. All five tumoroid models showed a significant decrease in AGCT viability at both 1 μM and 3 μM concentrations of abemaciclib compared to control (DMSO) (Fig. 4A). Dose-response experiments with abemaciclib were performed in two tumoroid models, with IC50 values of 0.89 μM and 11.7 μM (Fig. 4B).

Fig. 4.

Abemaciclib inhibits the growth of AGCT tumoroids. A, CCK-8 quantification of abemaciclib and DMSO (control) treated AGCT tumoroid GCT025, GCT027, GCT029, GCT031, and GCT032. Normalization was performed relative to the DMSO group, with DMSO value normalized to 1. Results of a two-sided t-test are shown. * p value ≤0.05; ** p value ≤0.01. B, Dose-response curve for abemaciclib treatment of AGCT tumoroid GCT028 and GCT030, 11 doses of abemaciclib used. Error bars, SD of the mean. IC50, Half-maximal inhibitory concentration.

Twenty-four SCID mice had their flanks injected with KGN tumor cells. Twelve days after injection, tumor volumes were measured, and mice were divided into two groups, treated with control or abemaciclib, with equal average tumor volume between groups. Mice were treated with daily oral gavage. Treatment lasted for 22 days. The average volume for the tumors at 22 days of treatment was 79mm³ for the abemaciclib group and 108mm³ for the control group, p = 0.003. The average mass of the tumors after excision was 26 mg in the abemaciclib group and 45 mg in the control group, p = 0.0002 (Fig. 5A). No adverse events were noted in the treated mice. No mouse lost more than 10 % of its starting body weight (Fig. 5B).

Fig. 5.

Abemaciclib inhibits growth of KGN cells injected in mice. A, Tumor mass in mice injected with KGN tumor cells and treated with control versus abemaciclib. Error bars, SD of the mean. Results of a two-sided t-test are shown. ** p value ≤0.01. B, Weights at different time points of mice treated with either abemaciclib or control. C, SF1 staining of vehicle control-treated xenograft tumor. D, SF1 staining of abemaciclib-treated xenograft tumor.

Six mouse tumors, three from the control group and three from the abemaciclib group, were stained for SF1, a marker for sex-cord stromal tumors. SF1 staining was seen in the spindled tumor cells, with nuclear localization in all 6 tumors tested. When compared to one another, the control cohort displayed a higher density of cells staining. The intensity of staining also showed differences, with the control cohort showing moderate to strong staining diffusely (Fig. 5C), while the treated cohort showed a weaker to moderate staining (Fig. 5D), indicating a treatment effect.

A retrospective review of the MD Anderson Rare Gynecologic Malignancy Registry identified 11 patients with recurrent AGCT treated with a CDK4/6 inhibitor. Before being initiated on CDK4/6 inhibitors, patients had a median of 5 lines of systemic treatment (range 2–13) and underwent a median of 4 cytoreductive surgeries (range 2–8) for recurrent AGCT. Nine patients (81.8 %) received palbociclib, and 2 (18.2 %) received abemaciclib. Nine patients (81.8 %) received a CDK4/6 inhibitor combined with endocrine therapy, the other two (18.2 %) received a CDK4/6 inhibitor alone. Best response by imaging for this cohort included partial response (3, 27.3 %), stable disease (6, 54.5 %,) and progressive disease (2, 18.2 %). The median duration of response was 6 months. Both drugs were generally well tolerated by patients. There were 2 cases of dose reduction reported in the palbociclib cohort due to recurrent neutropenia and 1 dose reduction in the abemaciclib cohort due to persistent diarrhea.

4. Conclusions

CDK4/6 inhibitors demonstrated high inhibition of cell proliferation in the KGN cell line and multiple AGCT tumoroid models. Xenograft models showed reduced size in tumor mass and differences in SF1 staining indicative of treatment effect, in the abemaciclib-treated cohort compared to controls. Despite a wide range of prior treatments (ranging from 0 to 5 lines), abemaciclib was effective across all tumoroid models tested. The addition of fulvestrant did not strengthen the potency of CDK4/6 inhibitors despite the presence of exogenous estrogen. Direct comparisons of abemaciclib efficacy across tumoroid models were limited by cell count and viability differences in each model. As a result, no conclusions regarding potential biomarkers for improved response can be drawn from this study.

CDK4/6 inhibitors have shown efficacy in small case series involving patients with AGCTs. In a case series of seven patients treated with palbociclib combined with either fulvestrant or letrozole, three patients achieved a partial response, and two had stable disease [35]. Another case series involving four patients treated with CDK4/6 inhibitors along-side endocrine therapy reported clinical benefit in three patients [36]. Both case series report that patients tolerated the combination well. Our cohort demonstrated similar results. Of the eleven patients with AGCTs treated with CDK4/6 inhibitors with or without endocrine therapy in our cohort, three patients had a partial response, and two had stable disease with a median response of 6 months. Combining these three groups, the overall response rate is 32 % (7/22), and the clinical benefit rate, defined as a partial response or stable disease >12 weeks, is 59 % (13/22) for CDK4/6 inhibitors in recurrent AGCTs. Currently, a phase II open-trial of abemaciclib and letrozole in patients with estrogen-receptor-positive rare ovarian cancer (ALEPRO trial) is ongoing [37].² Hopefully, this will further add to the data of CDK4/6 inhibitors in AGCTs.

The efficacy of CDK4/6 inhibitors in breast cancer and the associated markers of resistance have been well documented in the literature. Established markers of resistance to CDK4/6 inhibitors include loss of RB1, PTEN, and FAT1, as well as amplification of CDK2, CDKN family, Cyclin E1/E2, CDK6, AKT1, RAS, HER2, TK1, FGFR1/2, myc, p16, S6K1, and ERBB2. [38–54]. Additionally, loss of estrogen receptor expression has been associated with increased resistance [39,47]. However, none of the patients with somatic tumor testing in this cohort harbored the aforementioned resistance markers to CDK4/6 inhibitors. Furthermore, molecular profiling of 423 AGCTs revealed that, aside from the ubiquitous FOXL2 mutation, the most common pathogenic variants were TERT (present in 54 % of AGCTs) and KMT2D/LL2 (present in 17 % of AGCTs) [14]. TERT promoter mutations in recurrent AGCTs have been associated with less favorable outcomes [55], and KMT2D mutations have been associated with an increased risk of recurrence [56]. Furthermore, altered hormonal signaling has been noted in recurrent AGCTs compared to primary tumors [57], indicating that CDK4/6 inhibitors in combination with endocrine therapy may be particularly effective. Given the limited evidence of resistance linked to the most common molecular alterations in AGCTs and mutations associated with worse outcomes in AGCTs, these tumors may be promising candidates for CDK4/6 inhibitor therapy.

This preclinical data, including cell lines, tumoroid models, and xenograft models, supports the design of clinical trials to test the efficacy of CDK4/6 inhibitors in treating recurrent AGCT. Given the promising results across various model systems, these findings provide a strong rationale for translating this approach into clinical settings.

HIGHLIGHTS.

• CDK4/6 inhibitors show potent tumor cell proliferation inhibition in pre-clinical models of adult granulosa cell tumors.

• Xenografts showed a significant decrease in adult granulosa cell tumor volume and mass with abemaciclib compared to control.

• CDK4/6 should be explored further in adult granulosa cell tumors with clinical trials.