Phase II Trial of Pembrolizumab and Epacadostat in Recurrent Clear Cell Carcinoma of the Ovary: an NRG Oncology Study GY016

Abstract

INTRODUCTION:

Early reports of PD-1 inhibition in ovarian clear cell carcinomas (OCCC) demonstrate promising response. We evaluated the combination of pembrolizumab and IDO-1 inhibitor epacadostat in patients with recurrent OCCC.

METHODS:

This single arm, two-stage, phase 2 trial included those with measurable disease and 1–3 prior regimens. Patients received intravenous pembrolizumab 200 mg every 3 weeks and oral epacadostat 100 mg twice a day. Primary endpoint was overall response rate (ORR), secondary endpoints were toxicity, progression-free survival (PFS) and overall survival (OS). The study was powered to detect an absolute 25% increase in response (15% to 40%).

RESULTS:

Between September 28, 2018 and April 10, 2019, 14 patients enrolled at first stage. Rate of accrual was 2.3 patients per month. Median age was 65 years (44–89), 10 (71.4%) had ≥ 2 prior regimens. ORR was 21% (95% CI 5–51%) within 7 months of study entry with 3 partial responses, and 4 had stable disease (disease control rate 50%). Median PFS was 4.8 months (95% CI: 1.9–9.6), OS 18.9 months (95% CI: 1.9-NR). Most common grade ≥ 3 adverse events were electrolyte abnormalities and gastrointestinal pain, nausea, vomiting, bowel obstruction. In July 2019, the study reached the prespecified criteria to re-open to second stage; however, the study closed prematurely in February 2021 due to insufficient drug supply.

CONCLUSIONS:

Pembrolizumab and epacadostat demonstrated an ORR of 21% in this small cohort of recurrent OCCC. The rapid rate of accrual highlights the enthusiasm and need for therapeutic studies in patients with OCCC.

INTRODUCTION

Ovarian clear cell carcinoma (OCCC) is a rare histologic subtype accounting for <10% of epithelial ovarian cancers in North America and Europe [1]. OCCC has a worse prognosis and survival compared to high grade serous ovarian cancer with a response rate of <10% in recurrent disease with platinum-based chemotherapy [2]. There is an unmet need for effective therapies in this disease type, given the particularly poor prognosis of OCCC upon recurrence, the lack of effective treatment, and the rarity of this disease. Future treatment of this disease may rest on novel therapies.

Immune checkpoint blockade is increasingly being used as a therapeutic option in gynecologic malignancies. In epithelial ovarian cancer, response to single agent programmed cell death protein 1 (PD-1) or programmed death-ligand 1 (PD-L1) inhibitors has ranged from 8–13% [3,4] in studies that include all histologic subtypes. Notably, small studies have suggested that immune checkpoint inhibitors may be of particular benefit in those with clear cell histology. In a phase 2 study by Haminishi et al [4], 20 patients with heavily treated platinum-resistant ovarian cancer were given nivolumab. Two patients (10%) had a durable complete response (CR), one of whom had clear cell carcinoma with peritoneal dissemination which disappeared after two cycles of nivolumab. In a phase 1b study of avelumab in recurrent or refractory ovarian cancer, two patients with OCCC had partial responses [5]. In a phase 2 randomized trial of nivolumab versus nivolumab and ipilimumab for recurrent ovarian cancer, a subset analysis demonstrated those with clear cell histology were five times more likely to respond to nivolumab and ipilimumab compared to other histologic subtypes, though clear cell histology represented only 12% of the study cohort [6]. Although these early studies have shown promise, in order to improve upon these results, strategies incorporating combinations of immunotherapy are needed. Furthermore, with the small number of clear cell patients in these previous studies, histotype-specific trials specifically for patients with OCCC are also needed.

Indoleamine 2, 3-dioxygenase 1 (IDO1) is an enzyme that catalyzes oxidative catabolism of the amino acid tryptophan (Trp) to kynurenine [7], which results in the inhibition of antitumor cell–mediated immune responses. IDO1 is prevalent in approximately 56% of ovarian tumors and correlates with a reduced number of CD8+ tumor infiltrating lymphocytes (TILs) [8]. High IDO expression is associated with impaired overall and progression-free survival, with a trend across all histologic subtypes, including clear cell cancer [8].

Epacadostat (INCB024360) is a potent, selective oral inhibitor of IDO1. Inhibiting IDO1 leads to decreased kynurenine, and downstream downregulation of regulatory T-cells (Tregs), thereby increasing the antitumor immune response. Results from a phase I/II study of epacadostat in combination with MK-3475 (pembrolizumab), an anti-PD-1 antibody, were reported in patients with advanced cancers [9] with an overall response rate of 53% with 3 complete responses and a disease control rate of 74%. Promising anti-tumor activity was reported in patients with advanced renal cell carcinoma with response rates of 37%.

With poor outcomes related to recurrent OCCC, early studies demonstrating durable objective responses with PD-1 inhibition with this histologic subtype, and the need for a combination approach to improve response rates even further, we sought to examine the efficacy of pembrolizumab and epacadostat (IDO1 inhibitor) in recurrent OCCC.

METHODS

Patients

Eligibility criteria included patients with recurrent or persistent clear cell carcinoma of the ovary with histologic confirmation on pathology report of the primary tumor of at least 50% clear cell histomorphology; measurable disease according to RECIST, version 1.1; history of one primary platinum-based chemotherapy with up to two additional cytotoxic regimens for management of recurrent or persistent disease; and an Eastern Cooperative Oncology Group performance status score of 0 (fully active) or 1 (ambulatory, able to carry out light or sedentary work). Ineligibility criteria were those with prior immune checkpoint inhibitor therapy, history of autoimmune disease or those requiring immunosuppressive therapy.

Trial Design and Interventions

The study (NRG-GY016; ClinicalTrials.gov identifier: 03602586) was a single arm phase II trial. Eligible patients received pembrolizumab 200 mg intravenously every 3 weeks and epacadostat 100 mg orally twice a day. Treatment was continued until disease progression, or adverse effects prohibited further treatment.

Disease was assessed with imaging of the chest, abdomen, and pelvis according to RECIST, version 1.1; physical examination; and CA-125. Imaging was required 12 weeks after the first study treatment then every 6 weeks for 49 weeks and then every 12 weeks until disease progression or treatment discontinuation.

Patients were evaluated for adverse events according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE), version 5.0.

Endpoints and Statistical Analysis

The primary endpoint was objective tumor response (complete or partial) by RECIST, version 1.1, within 7 months of enrollment. The targeted sample size was 23. The study followed a flexible two-stage design [10] with early stopping for futility of the experimental regimen. The first stage of this study targeted an accrual of 12 patients, but permitted accrual to range from 10 to 14 patients. Accrual to the second stage was contingent on at least 2 responses, after which the study would reopen to enroll a total of 23 patients, with a permitted range from 21 to 25 patients. The study was powered to detect an absolute 25% increase in response, from a baseline of 15% to 40%. This provided a 90% probability of claiming the experimental agent as promising with a one-sided type I error rate of 0.10. The probability of early termination was 45% under the null hypothesis. All enrolled patients deemed eligible and who initiated study treatment were included in the analysis of objective response.

Secondary endpoints included toxicity, progression-free survival (PFS) and overall survival (OS). Toxicity was assessed in all eligible patients since all patients received at least one dose of treatment. Frequencies of adverse events were tabulated by term within system organ class. PFS was defined as duration of time from start of treatment until time of cancer progression demonstrated by imaging according to RECIST, version 1.1, symptomatic deterioration, or death from any cause. OS was defined as the duration of time from start of treatment to time of death or the date of last contact. PFS and OS were assessed in all eligible patients. The distributions of PFS and OS were estimated and graphed using the Kaplan-Meier product limit method.

Patient baseline characteristics including age category, ethnicity, race, prior radiation therapy, prior chemotherapy, prior bevacizumab, and mismatch repair deficiency were summarized as frequencies and percentages.

Study accrual in the first stage was completed. Although the study met criteria to open to second stage, the study was closed prematurely due to insufficient drug supply.

RESULTS

Patient and Trial Interventions

Between September 28, 2018 and April 10, 2019, 14 patients were enrolled in the United States during the first stage of accrual. The accrual rate in the first stage was 2.3 patients per month, which surpassed the projected accrual rate of 1.5 patients per month. There was a three-month suspension between the end of the first stage and ability to open at second stage. The study did not proceed to second stage due to insufficient drug supply.

Baseline clinical and tumor characteristics and prior treatment histories are described in Table 1. Among the 14 evaluable patients, median age was 65 years (range 44–89). All patients were white, with 1 (7.1%) patient of Hispanic ethnicity. All patients had prior chemotherapy; 4 patients (29%) received one regimen, 5 patients (36%) received two regimens and 5 patients (36%) received three regimens. Three patients (21.4%) had prior bevacizumab, and 3 patients (21.4%) had prior radiation therapy of which 2 had pelvic radiation. There were 5 patients whose tumors were not mismatch repair (MMR) deficient and 9 patients where MMR-status was unknown. There were 0 patients with a known MMR deficient tumor (Table 1).

Table 1.

Baseline patient and tumor characteristics

	Single Arm: Pembrolizumab and Epacadostat (n=14)
Age (years)  Median (min-max)  Q1-Q3  40–49  50–59  60–69  70–79  80–89	65 (44–89) 51–71 1 (7.1%) 4 (28.6%) 5 (35.7%) 3 (21.4%) 1 (7.1%)
Ethnicity  Hispanic or Latino  Non Hispanic or Latino	1 (7.1%) 13 (92.9%)
Race  White	14 (100%)
Prior Radiation Therapy  Yes  No	3 (21.4%) 11 (78.6%)
Prior Chemotherapy  1 regimen  2 regimens  3 regimens	4 (28.6%) 5 (35.7%) 5 (35.7%)
Prior Bevacizumab  Yes  No	3 (21.4%) 11 (78.6%)
Mismatch Repair Deficient  No  Unknown	5 (35.7%) 9 (64.3%)

All 14 patients initiated protocol therapy. Eleven patients (79%) came off treatment following disease progression. There were two patients (14%) who discontinued treatment due to adverse events, while one patient (7%) refused further protocol therapy after receiving one cycle, as she wished to stop any further cancer treatment. There were no patients on treatment at time of study lock.

Safety

Descriptive statistics for treatment-related adverse events are shown in Table 2. The most common adverse events were grade 1 and 2, consisting of fatigue, nausea, vomiting, anemia, constipation, and anorexia, reported in 50% or more of patients. Six patients (42.9%) had grade 3 adverse events, most commonly metabolism and nutritional disorders such as electrolyte abnormalities or anorexia, and gastrointestinal disorders such as abdominal pain, nausea, vomiting, bowel obstruction. Among those with gastrointestinal disorders there was one documented grade 2 colitis. There was one grade 4 serious adverse event (SAE), disseminated intravascular coagulation, which was deemed unlikely related to treatment. There were three grade 5 SAEs reported. Two deaths were not otherwise specified; one of which was deemed as probably treatment-related and definitely related to disease, and the second of which was deemed as unlikely related to treatment and definitely related to disease. The third death was due to intestinal perforation during cycle 1, which was deemed as unlikely treatment-related and probably related to disease.

Table 2.

Adverse Events, Regardless of Attribution, Occurring >10% Frequency

Adverse Event	N (%)
Blood and lymphatic system disorders  Anemia	7 (50)
Endocrine Disorders  Hyperthyroidism  Hypothyroidism	3 (21.4) 4 (28.6)
Gastrointestinal Disorders  Abdominal Pain  Constipation  Diarrhea  Dry Mouth  Mucositis Oral  Nausea  Small intestinal obstruction  Vomiting	4 (28.6) 8 (57.1) 4 (28.6) 3 (21.4) 2 (14.3) 9 (64.3) 3 (21.4) 7 (50)
General Disorders  Chills  Death NOS  Edema limbs  Fatigue  Fever  Pain	2 (14.3) 2 (14.3) 3 (21.4) 11 (78.6) 3 (21.4) 2 (14.3)
Infections and Infestations  Thrush	2 (14.3)
Metabolism and Nutrition Disorders  Anorexia  Hyperglycemia  Hypoalbuminemia  Hypocalcemia  Hypokalemia  Hypomagnesemia  Hyponatremia	7 (50) 5 (35.7) 3 (21.4) 2 (14.3) 3 (21.4) 2 (14.3) 3 (21.4)
Musculoskeletal and Connective Tissue Disorders  Myalgia	2 (14.3)
Nervous System Disorders  Dizziness  Headache  Other	2 (14.3) 2 (14.3) 2 (14.3)
Psychiatric Disorders  Insomnia	2 (14.3)
Respiratory and Thoracic Disorders  Cough  Dyspnea	4 (28.6) 3 (21.4)
Skin and Subcutaneous Disorders  Rash acneiform  Rash maculo-papular	2 (14.3) 4 (28.6)
Vascular Disorders  Hot flashes	2 (14.3)

Activity

Objective responses occurred in three patients in the first stage within 7 months of study entry (3 partial response (PR), 0 complete response (CR)), resulting in an overall response rate (ORR) of 21% (95% Exact CI 5%−51%). The median duration of response was 6.9 months. An additional 4 patients had stable disease (SD), resulting in a disease control rate (DCR) of 50% with a median duration of 6.6 months. Figures 1 and 2 demonstrate the best percentage change in tumor target lesions compared to baseline and response to treatment over time, respectively. Among those with partial response, one patient had a 68% decrease from baseline at 12 weeks, while 2 patients had a 48–50% decrease from baseline at 12 weeks (Figure 1). One with partial response was MMR-proficient, while the two others had unknown MMR status. Among those with response, 2 patients completed 33 cycles and 24 cycles of both drugs. One patient completed 14 cycles of pembrolizumab, but stopped the epacadostat after 1 cycle due to a grade 3 rash and symptoms of an allergic reaction. Length of treatment duration and responses are shown in Figure 2.

Figure 1.

Radiologic response to pembrolizumab and epacadostat in patients with recurrent ovarian clear cell carcinoma. Waterfall plot illustrating the best objective response in ten patients who had at least one post-treatment imaging assessment. Each bar represents a patient and shows the maximum percentage change from baseline. Patient 7 on the plot had a 0% change in response. Patient 9 received one cycle of epacadostat and 14 cycles of pembrolizumab.

Figure 2.

Swimmer’s plot illustrating treatment duration and durability of objective responses in patients on pembrolizumab and epacadostat for recurrent ovarian clear cell carcinoma. Each bar represents one subject.

At the time of analysis, all 14 patients either progressed or died, with 12 progressions and 2 deaths. Time from the start of study treatment to progression or death ranged from 0.4 months to 22.5 months. Median PFS was 4.8 months (95% CI 1.9–9.6) (Figure 3 and 4). Time from the start of study treatment to death or censoring ranged from 1.2 months to 36.3 months. Median OS was 18.9 months (95% CI 1.9-NR).

Figure 3.

Progression-Free Survival

Figure 4.

Overall Survival

In July 2019 the study met criteria to open to the second stage. However, at that point a phase 3 trial in patients with melanoma evaluating the combination of pembrolizumab and epacadostat was reported as failing to provide benefit [11]. Studies using this combination in other solid tumors subsequently closed. There was some discussion as to whether the dose of epacadostat at 100 mg twice a day was sufficient as subsequent pharmacodynamic data suggested a higher dose of 600 mg twice a day would be more effective in inhibiting IDO1. After holding for amendment, the study closed prematurely in February 2021 without opening to the second stage due to insufficient drug supply.

DISCUSSION

In this study, the combination of pembrolizumab and epacadostat had 3 partial responses with an overall response rate of 21% in the first 14 patients enrolled in the first stage, before the study closed due to insufficient drug supply. Unfortunately, it is difficult to conclude whether this is an effective combination for patients with OCCC given the underpowered sample size due to early closure.

Pembrolizumab as a monotherapy has also now been evaluated in OCCC. In an abstract presented at the European Society of Medical Oncology (ESMO) in 2022, Kristeleit et al. reported on the phase 2 PEACOCC trial [12]. There were 48 evaluable patients with recurrent clear cell gynecologic malignancies, 84% of which were ovarian cancers, treated with single agent pembrolizumab. The ORR was 25% (90% CI 15.1–37.3) with 1 CR and 11 PRs. Median PFS was 12.2 weeks (3 months) (95% CI 5.9–32.9) and median OS was 71 weeks (17.75 months) (95% CI 29.1–137.6). Tan et al. reported an abstract at the American Society of Clinical Oncology (ASCO) in 2022 of a multicenter phase 2 randomized trial of durvalumab vs physician choice chemotherapy for recurrent OCCC (MOCCA/APGOTOV2/GCGS-OV3) [13]. Nine academic centers from Singapore, South Korea, and Australia enrolled patients. In this 2:1 randomization, 31 patients received durvalumab, while 15 patients received physician choice chemotherapy. ORR was 10.7% in the durvalumab group, and 18.8% in the chemotherapy group (p=0.884). Median PFS was 7.4 weeks (1.85 months) in the durvalumab group vs 14 weeks (3.5 months) in the chemotherapy group (HR 1.5, 95% CI 0.8–2.8, p=0.89). There was no significant difference in ORR, PFS or clinical benefit rate between these two groups for recurrent OCCC.

It is hypothesized that combination immunotherapy may be more effective than monotherapy. However, it is difficult to elucidate the additional benefit of a second agent in single arm phase 2 studies. In the current study the combination of checkpoint inhibitor with IDO-1 inhibitor had an ORR of 21%. A retrospective case series reported on 16 patients with recurrent OCCC treated with PD-1/PDL-1 inhibitor +/− CTLA-4 blockade. Four patients (25%) had a durable clinical benefit, of which 3 had combination with CTLA-4 blockade [14]. Additionally, the phase 2 INOVA trial evaluates sintilimab (PD-1 inhibitor) and bevacizumab in recurrent OCCC. There is planned accrual of 38 patients, but the first 23 patients were presented as preliminary results at ESMO in 2022. ORR was 40% (8 patients) with a disease control rate of 75% [15]. This trial is continuing to enroll patients [16] with final results pending. Although the ideal study design would compare monotherapy checkpoint inhibitor vs combination therapy to see if an additional agent confers benefit, this requires accrual of a larger sample size, which can be challenging for studies with rare tumors.

Phase 1/2 results of pembrolizumab and epacadostat in advanced solid tumors (ECHO-202/KEYNOTE-037) demonstrated promising results [17], such that the combination of epacadostat 100 mg twice daily with pembrolizumab 200 mg once every 3 weeks was compared with placebo plus pembrolizumab in a large phase 3, ECHO-301/KEYNOTE-252 trial for advanced melanoma. However, this trial showed that epacadostat plus pembrolizumab did not improve PFS and OS when compared to pembrolizumab alone [11]. The failure of the ECHO-301/KEYNOTE-252 trial was accompanied by negative results from several phase 2 trials in other solid tumors such as non-small cell lung cancer, urothelial cancer, breast cancer and renal cell carcinoma. No apparent clinical benefit was observed in these trials. For this reason, the study did not proceed to second stage. Various hypotheses explaining these disappointing results include insufficient inhibition of IDO1 by the 100 mg dose of epacadostat, no selection of patients based on IDO1 expression, and inadequate blockade of the IDO1 downstream pathway [18,19]. It is unlikely that there will be future study of this drug combination in OCCC.

In addition to immune checkpoint inhibitors, another area of drug development in OCCC is to target those with ARID1A mutations. ARID1A mutations are detected in approximately 50% of OCCC [20]. Loss of ARID1A function is associated with increased microsatellite instability and tumor mutational burden, demonstrated by bioinformatic analysis in gastric cancer and OCCC [21–23]. Studies exploiting ARID1A synthetic lethal interactions such as ATR inhibition or EZH2 inhibition are ongoing [24,25]. Combinations of immunotherapy and agents targeting those with ARID1A mutations may be a novel path forward to explore in OCCC patients with both MMR-deficiency and ARID1A mutation.

Strengths of this study were the enrollment of patients with only ovarian clear cell histology, without the addition of other epithelial subtypes. There was a rapid accrual in the first stage which surpassed the estimated accrual rate, highlighting the opportunity to study this rare disease. Limitations are that the study was concluded early with insufficient power in the sample size to come to any definitive conclusions regarding the combination of pembrolizumab and epacadostat. The small sample size prevented any meaningful subgroup analyses such as the impact of different patient characteristics on patient response. The additional benefit of epacadostat to pembrolizumab alone is difficult to elucidate in this small single arm phase 2 study. Finally, premature closure of the study resulted in not being able to complete any biomarker analysis related to response to immunotherapy such as PDL-1 expression or tumor mutational burden.

The combination of pembrolizumab and epacadostat in recurrent and metastatic OCCC conferred an ORR of 21% but the study was terminated early after the first stage. In this rare tumor there continues to be few options for treatment at this stage of disease. Continued support to develop prospective multicenter studies in rare tumors such as recurrent OCCC are essential to improve patient outcomes.

Highlights.

• Pembrolizumab and epacadostat had an overall response rate of 21% in this cohort of recurrent ovarian clear cell carcinomas

• It is difficult to conclude whether this is an effective combination given the underpowered sample size due to early closure

• Rapid accrual in the first stage surpassed the estimated accrual rate, highlighting the need for trials for this rare tumor