Pembrolizumab Plus Radiation Fails to Improve Recurrence-Free Survival in Early-Stage High-Intermediate–Risk dMMR Endometrial Cancer

The addition of pembrolizumab (Keytruda) to adjuvant radiation therapy did not significantly improve the proportion of patients free from recurrence at 3 years compared with radiation alone in patients with newly diagnosed, early-stage, high-intermediate–risk, mismatch repair–deficient (dMMR) endometrioid endometrial cancer, according to findings from the phase 3 NRG-GY020 trial (NCT04214067) presented at the 2026 ESMO Congress.¹

Toxicity was higher with the combination, and investigators concluded that radiation alone remains the standard of care in this population.

At the primary analysis, 73% of patients in the radiation-alone arm (n = 56) and 77% of patients in the radiation plus pembrolizumab arm (n = 112) were recurrence free at 3 years (Z = 0.508; P = .3057). Among the 15 failures (27%) in the radiation-alone arm, 10 (18%) were due to recurrence, and 5 (9%) were due to unknown status, including 3 consent withdrawals and 2 instances of suspended follow-up. Among the 26 failures (23%) in the combination arm, 12 (11%) were due to recurrence, and 14 (13%) were due to unknown status, including 12 consent withdrawals and 2 instances of suspended follow-up.

"These data do not support the addition of pembrolizumab to radiation in a dMMR, high-intermediate–risk endometrial cancer group," said presenting author Floor J. Backes, MD, MBA, a professor and gynecologic oncologist at The Ohio State University James Comprehensive Cancer Center in Columbus. "So, radiation alone, whether that's vaginal brachytherapy with or without [external beam radiation therapy], remains the standard in this population."

Why was pembrolizumab evaluated in this setting?

Seventy-five percent of patients with endometrial cancer present with early-stage disease. Backes noted that, as of 2018, patients with high-intermediate–risk dMMR endometrial cancer were known to have an extravaginal recurrence rate of approximately 19% vs 4.5% for patients with mismatch repair–proficient disease, and data from the GOG 249 trial showed an 82% 3-year recurrence-free survival (RFS) rate and an 18% distant recurrence rate in high-intermediate–risk disease despite external beam radiation or chemotherapy. Single-agent pembrolizumab received FDA accelerated approval in 2017 for dMMR recurrent cancer.

"It was clear that novel treatment options were needed for the specific dMMR endometrial cancer subset that was at high-intermediate risk of recurrence," Backes said.

Ahead of the meeting, the trial was noted for restricting enrollment to patients with dMMR disease, in contrast with the phase 3 KEYNOTE-B21 trial (NCT04634877), which evaluated pembrolizumab in an all-comers population.²

How was NRG-GY020 designed?

Eligible patients had stage I/II endometrioid endometrial cancer meeting GOG 249 high-intermediate–risk criteria based on age, grade, depth of invasion, and lymphovascular space invasion (LVSI); dMMR status by immunohistochemistry and/or microsatellite instability on local testing; surgical staging with hysterectomy, bilateral salpingo-oophorectomy, and sentinel lymph node assessment with or without full lymphadenectomy; an ECOG performance status of 0 to 2; and no measurable disease outside the uterus.¹

Patients were randomly assigned 2:1 to receive vaginal brachytherapy and/or external beam radiation therapy with or without pembrolizumab, stratified by radiation type. Vaginal brachytherapy was the default; physician choice of external beam radiation was permitted only for patients with stage IB grade 3 or stage II disease. Radiation began 4 to 9 weeks after surgery. Pembrolizumab was initiated within 7 days before the start of radiation and given for up to 1 year, initially at 200 mg intravenously every 3 weeks; in September 2020, during the COVID-19 pandemic, dosing changed to 400 mg every 6 weeks, reducing the number of cycles from 17 to 9.

The primary end point was the percentage of patients who were recurrence free at 3 years per RECIST 1.1 criteria by investigator assessment, analyzed as a dichotomized end point. Patients who died, recurred, or came off study within 3 years for any reason, as well as patients who remained censored at less than 3 years, were considered treatment failures. The study had 89% power to detect a 15% increase in 3-year RFS, from 80% to 95%. Secondary end points included safety, recurrence patterns, RFS and overall survival (OS) at 5 years, and patient-reported outcomes and quality of life.

Between February 2020 and August 2022, 168 patients were enrolled at 75 sites. The median age was 66 years (range, 41-85), 76.8% of patients were White, 62.5% had stage IB disease, 17.3% had stage II disease, 48.8% had grade 2 tumors, 33.3% had grade 3 tumors, 57.5% had LVSI-positive disease, and 70.2% received vaginal brachytherapy alone. In the combination arm, 59.8% of patients completed treatment and 40.2% discontinued, including 23 patients who discontinued due to adverse effects (AEs); in the radiation-alone arm, 94.6% of patients completed treatment.

What did secondary end point data show?

At the August 27, 2026, data cutoff, with a median follow-up of 53.8 months, 21% of patients in the radiation-alone arm and 14% of patients in the combination arm had experienced recurrence or death. The RFS rates beyond 3 years were 81% and 88%, respectively (log-rank P = .113), and the Cox proportional hazards model yielded an HR of 0.632 (95% CI, 0.299-1.337; P = .115). OS data were not mature, with 9 events reported to date: 4 in the radiation-alone arm and 5 in the combination arm.

Because the number of patients with unknown 3-year status was relatively large and unbalanced between arms, investigators conducted a sensitivity analysis. Assuming the probability of success among unknown patients in the pembrolizumab arm ranged from 80% to 95%, with the radiation-alone arm held at 80%, the probability of reversing the study's conclusion ranged from 6.9% to 23%.

What was the safety profile of the combination?

In the safety population (radiation alone, n = 53; combination, n = 111), the most common treatment-emergent AEs in the combination arm vs the radiation-alone arm included fatigue (57.7% vs 35.8%), diarrhea (45.0% vs 39.6%), maculopapular rash (32.4% vs 3.8%), hypertension (25.2% vs 15.1%), decreased lymphocyte count (25.2% vs 3.8%), pruritus (23.4% vs 0%), and hypothyroidism (16.2% vs 0%). Grade 3 or higher differences were driven mostly by grade 3 lymphopenia and hypertension; grade 3/4 decreased lymphocyte count occurred in 15.3% of patients in the combination arm, and grade 3/4 hypertension occurred in 11.7% vs 7.5% of patients, respectively.

Most immune-related AEs were grade 1 or 2. Grade 3/4 immune-related AEs in the combination arm included adrenal insufficiency (3.6%), maculopapular rash (1.8%), and hypophysitis (0.9%), as well as low rates of colitis and pancreatitis.

Backes noted that the high rate of failures for reasons other than recurrence, the relatively low overall event rate, and the limited sample size were limitations of the study. "There was a relatively low event rate, in total 13% [at] 3 years, with a limited sample size that reduced the ability to detect treatment differences when adding pembrolizumab to radiation," she said.

References

1. Backes FJ, et al. NRG-GY020: phase III randomized trial of radiation +/- MK-3475 (pembrolizumab) for newly diagnosed early-stage high intermediate risk (HIR) mismatch repair deficient (dMMR) endometrioid endometrial cancer (EC). Presented at: 2026 ESMO Congress; October 23-27, 2026; Madrid, Spain. Abstract 1239O.
2. Van Gorp T, Cibula D, Lv W, et al. ENGOT-en11/GOG-3053/KEYNOTE-B21: a randomised, double-blind, phase III study of pembrolizumab or placebo plus adjuvant chemotherapy with or without radiotherapy in patients with newly diagnosed, high-risk endometrial cancer. Ann Oncol. 2024;35(11):968-980. doi:10.1016/j.annonc.2024.08.2242