DART/SWOG/NCI phase II anti-CTLA-4/PD-1 trial: clear cell carcinomas of ovary, endometrium, cervix.

The trial, which took place at 1016 sites throughout the USA, was supervised and steered by the Early Therapeutics and Rare Cancer Committee of the SWOG Cancer Research Network/National Cancer Institute (NCI). Nivolumab and ipilimumab agents were provided by the NCI’s Cancer Therapy Evaluation Program (CTEP), under the NCI CRADA agreement with Bristol Myers Squibb (BMS). Every participant of the study freely provided a signed informed consent document, which was sanctioned by the human subject protection committee of each participating institution. All research was performed in accordance with relevant guidelines/regulations. This basket trial was designed to incorporate patients with rare and ultra-rare tumor histology with no ongoing clinical trial exploring dual immunotherapy. Rare cancers were specified as those with a yearly incidence of fewer than 6 in every 100 000 cases. The pathology and grading of the tumors were evaluated by the pathologists at the participating institutions or by local pathologists, with further review of the pathology reports by the main investigators of the study. The types of tumors were identified according to the criteria laid out in the fifth edition of the WHO Classification of Female Genital Tumors. No central pathology review was executed. This paper presents the findings from the cohorts of ovarian (cohort 46), endometrial (cohort 45), and cervical (cohort 42) CCCs. Patients eligible for the trial had a histologically confirmed diagnosis of clear cell gynecologic cancers (specifically ovarian, endometrial, and cervical CCCs) and either no other treatment options known to extend OS, declined other treatment, or had contraindications to other treatment. As the National Comprehensive Cancer Network advocates that the optimal approach for managing any patient with cancer is participation in a clinical trial, this trial aligns with their recommendation. Patients were initially classified at the discretion of the primary investigator at the site. Later, these classifications were reviewed and, for some patients, revised at the closure of the study by the authors, based on the pathology reports and clinical history. At enrollment, patients had to be 18 years or older, have a Zubrod performance status ranging from 0 to 2, and show adequate hematologic, hepatic, thyroid, adrenal axis, and renal function. Specific benchmarks for these functions included an absolute neutrophil count of at least 1000/µL, platelets of at least 75 000/µL, hemoglobin of at least 80 g/L, creatinine clearance of at least 50 mL/min, total bilirubin less than or equal to 2.0 times the institutional upper limit of normal (IULN), aspartate aminotransferase (AST) and alanine aminotransferase (ALT) less than or equal to 3.0 times IULN, thyroid stimulating hormone or free T4 serum less than or equal to IULN, and normal adrenocorticotropic hormone (ACTH). All participants were required to use effective contraception during the protocol, and all women capable of becoming pregnant needed to have a negative serum pregnancy test at the time of enrollment. Patients received intravenous treatment with nivolumab (240 mg every 2 weeks) and ipilimumab (1 mg/kg every 6 weeks) on a continual basis, with dose adjustments and temporary pauses in therapy specified in the protocol for handling treatment-related toxicities. The protocol is available in online supplemental file 1 . Patients were removed from protocol treatment for reasons, including disease progression, symptomatic worsening, any treatment delay exceeding 56 days, unacceptable or immune-related toxicity preventing the reduction of prednisone to below 10 mg daily, or at the patient’s request. At the start of each cycle (or at least every 6 weeks), patients were evaluated with a medical history, physical examination, laboratory analyses (complete blood count, comprehensive metabolic panel, thyroid-stimulating hormone, free thyroxine, ACTH, cortisol, lipase), and toxicity assessment. Changes in dosage for managing immune-related adverse events (irAEs) were made according to the specific guidance criteria provided. The burden of the disease was measured with imaging studies before the study, at weeks 8, 16, and 24, and then every 12 weeks until progression. The primary endpoint was ORR (confirmed CRs and partial responses (PRs)) as per Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1 criteria per investigator, and the study was powered to differentiate between an actual ORR of 5% (null hypothesis) vs 30% (alternative hypothesis). A two-stage design was used; if one or more of the first six patients had a confirmed CR or PR, an additional 10 patients were to be enrolled. Two or more patients with a confirmed CR or PR of 16 were considered evidence of activity (87% power, one-sided alpha=13%). Secondary objectives included PFS per RECIST V.1.1, OS, CBR (stable disease (SD) ≥6 months plus ORR), ORR per immune-related RECIST (iRECIST), PFS per iRECIST (iPFS), and toxicity assessment. PFS was measured from the first day of protocol treatment to the time of progression or death from any cause, with patients last known to be alive without progression censored at the date of last contact. OS was measured from the date of registration to the date of death from any cause, with patients last known to be alive censored at the date of last contact. PFS and OS were estimated with the Kaplan-Meier method; medians were computed using the Brookmeyer and Crowley method. The association between PFS and number of prior therapies was evaluated using a Cox regression model. Point estimate CIs (eg, 6-month PFS) were calculated using the log-log transformation. All analyses were executed using R V.4.3.3.

Between January 2017 and March 2023, patients were enrolled in the S1609 study, with the longest follow-up duration for any participant being 5 years. Altogether, 32 patients with gynecologic CCC were evaluable from 23 of the 1016 participating National Clinical Trial Network institutions. This included 19 evaluable patients with ovarian CCC, 8 evaluable patients with endometrial CCC, and 5 evaluable patients with cervical CCC. The median age for all evaluable patients was 57 years (range, 21–74 years). The number of prior therapies ranged from 1 to 8, among whom 3 had been previously treated with a PD-1 inhibitor. Full demographic information for all patients is outlined in table 1 . A table of representativeness of study participants is available in online supplemental table 1 . Clinical benefit=SD ≥6 months or confirmed objective responses. CR, complete response; DART, Dual Anti-CTLA-4 & Anti-PD-1 blockade in Rare Tumors; PR, partial response; RECIST, Response Evaluation Criteria in Solid Tumors; SD, stable disease. The ORR by RECIST criteria was 9.38% (3/32) overall for all 32 evaluable patients with gynecologic CCC. This included two CRs and one PR in three patients with ovarian CCC. The ORR for all patients increased to 12.5% when including 1 PR by iRECIST criteria confirmed on next imaging for a patient with cervical CCC lasting 26 months, with an OS of 32.0 months ( figures1 3 , online supplemental table 2 ). Among the two patients with ovarian CCC with a CR, there was a confirmed 100% regression ongoing at 47.8+ months and 40.5+ months, respectively. The CBR was 21.88% overall for all 32 (7/32) evaluable patients with gynecologic CCC. This included two CR, one PR, and two patients with SD >6 months with ovarian CCC and one PR by iRECIST and one SD >6 months in two patients with cervical CCC. PFS for the 7 patients with CBR was 63.6+, 47.8+, 40.5+, 50.8+, 7.4, 26, and 58.1+ months. Median OS was 21.7 months for all 32 evaluable patients. One patient with endometrial CCC had unconfirmed PR, while there were two unconfirmed PRs and one unconfirmed CR in patients with ovarian CCC, which were not included in the calculated ORR for all patients. Full outcomes are outlined in table 1 and figures1 3 (see also online supplemental tables 2–4 for patient-level data). There was not a significant association between number of prior systemic therapies and PFS (HR=1.02, 95% CI 0.78 to 1.23, p=0.88). Next-generation sequencing (NGS) was not included in the study protocol; however, available NGS data for some patients are provided in online supplemental table 5 . Notably, among these, one ovarian cancer patient who achieved a CR was MMR proficient and had PD-L1 expression of 1%–4%. Taken together, 17 of 32 patients (53%) experienced >grade 3 adverse events (AEs) possibly related to treatment; 7 of 32 patients (22%) discontinued therapy because of toxicity, and there were no treatment-related deaths. Some of the most common all-grade potential treatment-related AEs were fatigue, nausea, anorexia, pruritus, liver and thyroid dysfunction; among the most common grade 3–4 AEs possibly related to treatment were anemia (n=3/32), liver enzyme increased (n=8/32), and nausea (n=2/32) ( table 2 ). Due to this observed level of high-grade toxicity, modified dosing could be considered in future studies, particularly for patients undergoing therapy for extended periods of time. Online supplemental table 6 shows potential drug-related AEs of 32 evaluable patients with clear cell gynecologic cancers, stratified by ovarian, endometrial, and cervical clear cell cancers. DART, Dual Anti-CTLA-4 & Anti-PD-1 blockade in Rare Tumors.

Clear cell carcinoma (CCC) is a rare and aggressive form of cancer that typically arises in the female reproductive organs and makes up approximately 1% of all gynecological cancers. 1 In general, clear cell pathology is associated with worse outcomes and a more aggressive course compared with other pathologies for patients with advanced gynecologic cancers. 1 CCC of the ovary represents approximately 5% of epithelial ovarian cancer cases, and there is a disproportionately high incidence among women of East Asian heritage. 2 CCC of the endometrium accounts for only 2%–4% of cases and is a heterogeneous disease. 3 Endometrial CCC encompasses different molecular subgroups including POLE mutated (mut), mismatch repair deficient (MMRd), p53abn, and no specific molecular profile (NSMP). 3 A meta-analysis found that POLEmut, MMRd, p53abn, and NSMP accounted for about 2%, 10%, 30%, and 54% of patients with endometrial CCC. 3 CCCs of the cervix constitute 4% of cervical adenocarcinoma. 4 Cervical CCC was historically associated with in utero exposure to diethylstilbestrol but also reported without this exposure. 4 Treatment for gynecological CCC is based mainly on experience with non-clear cell adenocarcinomas. Compared with other gynecologic histologies, CCC is intrinsically less responsive to chemotherapy, including platinum-based regimens, which are the cornerstone of systemic treatment, highlighting the need to explore novel therapeutic approaches. 5 In various malignancies, such as renal and lung cancer, dual checkpoint inhibition using anti-programmed death-1 (PD-1) or its ligand (PD-L1) and anti-cytotoxic T lymphocyte antigen-4 (CTLA-4) checkpoint inhibitors has proven to be effective. 6 7 Immunotherapy using single-agent anti-PD1/PD-L1 agents has shown modest activity in patients with certain gynecologic cancers, such as cervical cancer and microsatellite-unstable endometrial cancer. 8 9 The biologic rationale to use combined immunotherapy in CCC of gynecological origin stems from pivotal clinical trials of anti-PD1/PD-L1 monotherapy in this cancer type. In a phase II clinical trial (MOCCA), durvalumab was compared with physician’s choice chemotherapy (PCC) in patients with recurrent ovarian CCC. No significant differences in progression-free survival (PFS), overall response rate (ORR), or clinical benefit rate (CBR) were observed between durvalumab and PCC. 10 Retrospectively, it was seen that those patients with a favorable response to durvalumab exhibited increased levels of circulating T-cells, B-cells, NK-cells, and specific cytokines. 11 In a phase II clinical trial (PEACOCC) of pembrolizumab in heavily pretreated advanced CCC of gynecological origin (ovarian, n=41; endometrial, n=6; cervical, n=1), pembrolizumab showed clinical benefit in a predominantly MMR-proficient population, with a tolerable safety profile. 12 The trial reported a 25% ORR, with a median duration of response of 13.1 months and median overall survival (OS) of 14.8 months. 12 A single-arm phase II clinical trial (INOVA) investigated sintilimab in combination with bevacizumab in 37 patients with relapsed or persistent ovarian CCC and previous exposure to at least one cycle of platinum-containing chemotherapy. 13 An ORR of 40.5%, including 5 complete responses (CRs), was demonstrated. The SWOG S1609 Dual Anti-CTLA-4 & Anti-PD-1 blockade in Rare Tumors (DART) trial represents the first combination immunotherapy prospective trial to include a cohort dedicated to CCC of gynecological origin, aiming to evaluate the safety and activity of this approach in patients with this rare cancer type.

The NCI/SWOG S1609 DART trial conducted by the Early Therapeutics and Rare Cancers Committee represents the first prospective trial to evaluate the efficacy of combination immunotherapy using ipilimumab and nivolumab in a cohort dedicated to rare gynecologic CCCs. The treatment regimen showed durable responses, with the greatest clinical activity observed in patients with CCCs of ovarian origin. Among 19 patients with ovarian CCCs, there were two CRs and one PR, yielding an ORR of 15.8%. In the cervical CCC cohort (n=5), one patient achieved SD for over 6 months (ongoing at 58+ months), and another had a PR by iRECIST criteria lasting 26 months. No clinical benefit was observed among patients with endometrial CCC. While responses appeared limited in endometrial and cervical CCC, the small sample sizes preclude definitive conclusions. Larger studies are needed to ascertain whether response rates differ meaningfully by primary site. Ovarian CCC may show greater immunotherapy responsiveness than endometrial or cervical CCC due to distinct biological features. Ovarian CCC more commonly harbors ARID1A mutations and PI3K/AKT/mTOR pathway alterations and often may have an association with endometriosis, changes associated with increased immunogenicity. 14 15 In contrast, endometrial and cervical CCC display more heterogeneous molecular profiles with fewer of these immune-favorable alterations, which may contribute to their comparatively lower response rates. 16 The durability of responses is noteworthy: two patients with ovarian CCCs achieved CRs lasting over 3 years, and five patients among the entire cohort of 32 patients have PFS ongoing beyond 3.5 to over 5 years. The findings of the current study are consistent with other trials evaluating combination immunotherapy in patients with gynecologic CCC. A recently presented phase II trial (BrUOG 354) compared nivolumab monotherapy to nivolumab plus ipilimumab in 44 patients with relapsed extra-renal CCC, including 36 with ovarian origin. 17 Combination therapy demonstrated a higher ORR (33% vs 14.3%) and improved median PFS (5.6 vs 2.2 months) and OS (24.6 vs 17 months), supporting its potential superiority over monotherapy, particularly in chemotherapy-resistant ovarian CCC. Similarly, the MoST-CIRCUIT trial 18 (2024) reported an ORR of 50% in patients with CCC of ovarian (n=24) and uterine (n=4) origin treated with nivolumab and ipilimumab. In this study, tumor mutation burden was significantly associated with response (3.1 vs 1.15 mutations/megabase; Mann-Whitney p=0.03). Collectively, data from these combination trials and our study suggest that dual immune checkpoint blockade provides durable antitumor activity in gynecologic CCCs, particularly ovarian CCC. However, predictive biomarkers are needed to better identify patients most likely to benefit. The BrUOG 354 17 , MoST-CIRCUIT 18 , and current DART trials reported varying ORRs. The ORR in the current study, including iRECIST criteria, was 12.5%, compared with 33.3% in BrUOG 354 and 50% in MoST-CIRCUIT. Notably, the MoST-CIRCUIT trial employed a more intensive dosing regimen—nivolumab 3 mg/kg and ipilimumab 1 mg/kg every 3 weeks with maintenance nivolumab—which may have contributed to the higher ORR, whereas the BrUOG 354 and DART studies used identical, less intensive dosing schedules. Differences in prior lines of therapy may also have influenced outcomes. The median number of prior treatments was one in both BrUOG 354 and MoST-CIRCUIT, while most patients in the DART trial had received multiple prior therapies (range: 1–8). Additionally, BrUOG 354 excluded patients with prior immunotherapy, whereas three patients in the DART cohort had previously received PD-L1 monotherapy. Although limited in number, this prior exposure could have affected response to subsequent combination immunotherapy. Other factors—such as small sample sizes, differences in inclusion criteria, and patient demographics—may also contribute to the variation in response rates across trials. The combination regimen demonstrated a toxicity profile consistent with prior experience using nivolumab and ipilimumab. In this study, the dose of ipilimumab was reduced to 1 mg/kg intravenously every 3 weeks, compared with the conventional 3 mg/kg dosing typically administered for up to four doses. Despite this lower dose, treatment-related toxicity remained significant: 17 of 32 patients (53.1%) experienced grade ≥3 AEs possibly related to therapy, and 7 patients (21.9%) discontinued treatment due to toxicity. No treatment-related deaths occurred. When comparing toxicity findings across combination immunotherapy studies in gynecologic CCC, the BrUOG 354 trial 17 reported serious treatment-related AEs in 14 patients (47%), including two cases of grade 4 pancreatic enzyme elevations. In the MoST-CIRCUIT trial, 18 which used a more intensive dosing regimen, 7 patients (25%) experienced grade 3/4 irAEs, and one patient died from treatment-related myocarditis. These findings underscore the importance of balancing efficacy and toxicity when selecting combination immunotherapy regimens in clinical practice and suggest that dose optimization warrants further investigation. Emerging data also indicate that combining immunotherapy with anti-angiogenic agents may offer improved tolerability without compromising efficacy. For instance, in the INOVA trial (ORR 40.5%), relapsed or persistent ovarian CCC patients treated with sintilimab and bevacizumab experienced only three grade 3 treatment-related AEs (7%), with no events ≥grade 4. 13 Similarly, in the LARA trial (ORR at 24 weeks of 26.7%), patients with recurrent, MMR-proficient gynecologic CCC received pembrolizumab and lenvatinib with improved tolerability; only 7.4% discontinued treatment due to toxicity. 19 Common grade ≥3 treatment-related AEs (≥5%) included hypertension (22.2%), AST elevation (11.1%), ALT elevation (7.4%), and thrombocytopenia (7.4%). Strengths of this study include the successful and timely enrollment of a diverse cohort of patients with rare gynecologic CCCs, including ultra-rare subtypes, from both academic and community practices across the United States (1016 participating sites), as well as the support of the NCI and SWOG. Limitations include the single-arm, non-randomized design, small sample size, heterogeneous patient population, and incomplete biomarker data, which limited the current subgroup analysis. The biomarker information is incomplete as this collection is a mandatory inclusion criterion in the trial protocol. A significant limitation of this study is the absence of a centralized pathology review. CCCs of the gynecologic tract are known for their histologic heterogeneity and diagnostic overlap with other high-grade müllerian tumors, which can lead to interobserver variability among pathologists. Without centralized confirmation, the potential for misclassification cannot be excluded. Another notable limitation of this study is the absence of MMR status and tumor mutational burden (TMB). Prior studies in clear cell ovarian carcinoma have suggested that MMR deficiency and elevated TMB may correlate with increased responsiveness to immune checkpoint inhibition. 20 21 Without systematic molecular profiling in our cohort, we were unable to determine whether biomarker-enriched subsets may have contributed to the observed responses. Notably, the majority of patients (59%) had tumors of ovarian origin, and only three had prior exposure to PD-1 inhibitors. As PD-1 inhibitors have now moved into front-line treatment for endometrial and cervical cancers, the applicability of this combination therapy—and its observed efficacy—may be influenced by evolving treatment landscapes.

In summary, ipilimumab plus nivolumab demonstrated clinical activity in gynecologic CCCs, with the greatest benefit observed in the ovarian subtype. Although small subgroup sizes limited statistical comparisons, two patients with ovarian CCC achieved CRs ongoing beyond 3 years. Across the full cohort of 32 patients, five experienced PFS exceeding 3.5 to over 5 years, and the CBR was 21.9% (7/32). Patients received standard-dose nivolumab (240 mg every 2 weeks) and reduced-dose ipilimumab (1 mg/kg every 6 weeks) continuously; however, toxicity remained a concern, with 22% (7/32) discontinuing treatment due to AEs. Correlative studies are underway to identify biomarkers of sensitivity and resistance to immune checkpoint inhibitors. Further prospective trials—with potential dosing modifications—are warranted to optimize combination immunotherapy strategies for this rare and challenging cancer subtype. CCC is a rare and aggressive malignancy, accounting for approximately 1% of gynecologic cancers. While dual immune checkpoint blockade with anti-CTLA-4 and anti-PD-1 inhibitors has shown efficacy across multiple tumor types, their activity in gynecologic CCC remains largely unexplored. In this cohort of 32 patients, the ORR was 9.4% by RECIST and 12.5% by immune-RECIST, including two ongoing CRs (both ovarian in origin), one PR, and one additional PR by immune-RECIST. The CBR was 21.9% (7/32), with PFS for these patients ranging from 7.4 to over 63 months. Median OS for the full cohort was 21.7 months. Treatment was discontinued due to toxicity in 21.9% of patients (7/32), though no treatment-related deaths occurred. These findings suggest that dual checkpoint blockade may be clinically active in a subset of patients with gynecologic CCC, particularly those with ovarian origin, with some achieving durable and CRs.