Pembrolizumab/Chemoradiation Prolongs Survival in Advanced Cervical Cancer

“These data are consistent with the prior interim analyses results and provide further support for pembrolizumab plus CCRT as the new standard of care for this population,” according to study author Linda R. Duska, MD, MPH.

Sustained survival improvements were observed in patients with newly diagnosed, high-risk, previously untreated advanced cervical cancer who received pembrolizumab (Keytruda) plus concurrent chemoradiotherapy (CCRT) vs those who received CRT only, according to a presentation on final analysis data from the phase 3 ENGOT-cx11/GOG-3407/KEYNOTE-A18 trial (NCT04221945) at the 2025 American Society of Clinical Oncology (ASCO) Annual Meeting.

At a median follow-up of 41.9 months, the hazard ratio for progression-free survival (PFS) was 0.72 (95% CI, 0.59-0.87). The 24-month PFS rates were 70.6% and 59.7% in the respective arms, and the 36-month PFS rates were 64.3% and 55.6%. The hazard ratio for overall survival (OS) was 0.73 (95% CI, 0.57-0.94). The 36-month OS rate in the pembrolizumab arm was 81.8% vs 74.4% in the placebo arm; the respective 48-month OS rates were 75.4% and 70.2%.

Post-progression therapy, which is known to affect OS, according to lead study author Linda R. Duska, MD, MPH, of the University of Virginia Health in Charlottesville, VA, increased since interim analysis 2 of the trial. At the time of the final analysis, 33.3% of patients had received post-progression immunotherapy.

“These data are consistent with the prior interim analyses results and provide further support for pembrolizumab plus CCRT as the new standard of care for this population,” Duska said in a late-breaking oral presentation of the data.

Keeping Up With KEYNOTE-A18: Eligibility, Treatment, Objectives

The randomized, double-blind, phase 3 study enrolled patients with high-risk locally advanced cervical cancer, which was defined as FIGO 2014 stage IB2 to IIB, node-positive disease, or FIGO 2014 stage III to IVA disease, irrespective of lymph node status. Patients were naive to treatment and had measurable or nonmeasurable disease by RECIST 1.1 criteria.

Study participants (n = 1060) were randomly assigned 1:1 to the pembrolizumab arm or the placebo arm. Patients received 40 mg/m2 of cisplatin once weekly for 5 cycles plus external beam radiotherapy (EBRT) followed by brachytherapy plus 200 mg of pembrolizumab or placebo every 3 weeks for 5 cycles followed by 400 mg of pembrolizumab or placebo every 6 weeks for 15 cycles. Stratification factors included type of EBRT that was planned (IMRT vs VMAT vs non-IMRT vs non-VMAT), disease stage at screening (IB2 to IIB N+ vs III to IVA), and planned total radiotherapy dose (<70 Gy vs ≥70 Gy).

The study had 2 primary end points: PFS by RECIST 1.1 criteria and OS. “These end points were analyzed in a hierarchical way, first PFS and then if PFS was significant, all of the alpha was recycled to OS,” Duska explained. Secondary end points included 24-month PFS rate, 36-month OS rate, objective response rate, patient-reported health-related quality of life, and safety.

The study had 2 planned interim analyses. The first had a data cutoff of January 2023 and a median follow-up of 18 months; it evaluated whether the pembrolizumab approach significantly improved PFS and OS over placebo. The second had a data cutoff of January 2024 and a median follow-up of 30 months; it further assessed whether the experimental arm significantly improved OS. The final analysis, data of which were shared at the meeting, had a data cutoff of January 2025 and a median follow-up of 42 months. “This analysis was designed as a final descriptive OS analysis with no hypothesis testing for PFS and OS,” Duska said.

A total of 1562 patients underwent screening and 1060 were randomly assigned; 529 were allocated to the pembrolizumab plus CCRT arm and 531 were allocated to the placebo plus CCRT arm; 528 and 530 patients, respectively, received treatment. In the pembrolizumab arm, 55.5% had completed treatment, 0.9% remained on treatment, and 43.6% discontinued treatment. The most common reason for discontinuation was radiographic progression (n = 109), followed by toxicity (n = 66) and participant decision (n = 31).

“The baseline characteristics were well balanced between the 2 arms,” Duska said. With regard to PD-L1 combined positive score (CPS) status, in the pembrolizumab arm, 94.9% had a status of 1 or higher and 4.2% had a status under 1; this information was missing for 0.9% of patients. In the placebo arm, 93.8% had a status of 1 or higher and 5.3% had a status of under 1; this information was missing for 0.9% of patients. Slightly more than half of patients in both arms had stage III to IVA disease (56.0% vs 57.4%). Most patients in the experimental and control arms were planned to receive IMRT or VMAT (88.7% vs 88.5%) at a dose of 70 Gy or higher (91.1% vs 91.3%).

Treatment exposure was also well balanced between the groups, according to Duska. The mean number of cycles of pembrolizumab or placebo was 20 in both arms; for cisplatin, it was 5 cycles for both arms. The median overall treatment time for radiation therapy was 52 days in both arms and almost 75% of patients in both arms completed it within 56 days (74.2% vs 74.7%). The median duration of EBRT was 37 days in both arms and the median duration of brachytherapy was 12 days in both arms.

With regard to radiation exposure, the high-risk clinical target volume (HR-CTV) total EQD2 dose was 87 Gy in both arms. Most patients received IMRT or VMAT (89.4% vs 87.5%). The total EQD2 dose to the elective target volume and to the lymph nodes was 44 and 56, respectively, in both arms. With regard to brachytherapy, 3D HR-CTV D90% was approximately 88% in both arms (88.4% vs 86.8%); 94.3% of patients in both arms received high-dose rate brachytherapy with a total EQD2 dose of 41.

Comparing the Data That Came Before With the Current Analysis

Findings from interim analysis 1 showed that a median follow-up of 17.9 months, the study met its PFS end point, demonstrating a statistically significant improvement in PFS with pembrolizumab vs placebo, with a hazard ratio of 0.70 (95% CI, 0.55-0.89; P = .0020). The 24-month PFS rates in the respective arms were 67.8% and 57.3%. At this early analysis, the percentages of patients censored in the pembrolizumab and placebo arms were 78.3% and 71.0%, with an 88.5% information fraction. “Now…the censoring has decreased, and the hazard ratio is very similar, at 0.72. You also see the continued separation of the curves at 24 months, continuing to 36 months, and continuing beyond 36 months,” Duska noted. “The data at the end of the curve, all the way at the end of the graph, represents very few patients due to an increase in censoring.”

The trial’s OS end point was met at the time of interim analysis 2, which showed that at a median follow-up of 29.9 months, the hazard ratio for OS was 0.67 (95% CI, 0.50-0.90; P = .0040), which was determined to be statistically significant. At 36 months, 82.6% of those on the pembrolizumab arm were still alive vs 74.8% of those on the placebo arm. Duska noted significant censoring at this relatively early time point, with an information fraction of 76.7%.

At the final analysis, “with decreased censoring, we see a very similar hazard ratio, and we also see almost identical 36-month survival numbers,” Duska explained. “You can also see that compared to the prior curve, where the censoring really started increasing at around 12 months, we’ve now moved to the censoring increasing significantly at the 24-month time point…You can see almost identical 36-month survival in both analyses.”

Subgroup Analyses

Data from subgroup analyses reported at the final analysis were comparable to prior data, which showed that OS in all protocol-specified subgroups favored pembrolizumab over placebo with the exception of those aged 65 years or older (HR, 1.18; 95% CI, 0.58-2.43).

“I want to draw your attention to the FIGO 2014 stage IB2 to IIB [subgroup],” Duska noted. “Here, the hazard ratio was 0.92, which is very similar to prior subgroup analyses. I also want to point out that in this subgroup, there were very few events, at only 21%.”

Safety Spotlight

The safety profile of pembrolizumab plus CCRT proved to be manageable and in line with the known profiles of the individual agents. No new safety signals were observed with longer follow-up.

Treatment-related adverse effects (TRAEs) occurred in 97.0% of those in the pembrolizumab arm vs 96.8% of those in the placebo arm; they were grade 3 or higher for 69.5% and 61.5% of cases, respectively. Serious TRAEs occurred in 19.7% and 13.6% of patients, respectively. TRAEs led to discontinuation of any treatment for 18.9% and 13.0% of patients, respectively, and proved fatal for 2 patients in each arm.

Immune-mediated AEs occurred in 39.8% of those who received pembrolizumab vs 17.5% of those given placebo; they were grade 3 or higher for 5.1% and 1.3% of patients. Serious immune-mediated AEs occurred in 4.0% and 1.1% of patients, respectively. Immune-mediated AEs led to discontinuation of any treatment for 3.0% of those in the pembrolizumab arm vs 0.8% of those in the placebo arm; they proved fatal for 1 patient in the pembrolizumab arm.

The most common immune-mediated AEs reported in the pembrolizumab and placebo arms, respectively, included hypothyroidism (22.9% vs 7.4%), hyperthyroidism (12.1% vs 2.8%), gastritis (4.5% vs 4.0%), colitis (3.0% vs 2.1%), thyroiditis (2.3% vs 0.4%), pneumonitis (1.9% vs 0.9%), severe skin reactions (1.5% vs 0.9%), adrenal insufficiency (0.9% vs 0.2%), nephritis (0.6% vs 0.2%), and pancreatitis (0.6% vs 0.2%).

Disclosures: Duska disclosed serving in a consulting or advisory role for Ellipses Pharma (Inst), Inovio Pharmaceuticals, Merk (Inst), and Regeneron. Research funding was received from Abbvie (Inst), Aduro Biotech (Inst), Advaxis (Inst), Aeterna Zentaris (Inst), Bristol Myers Squibb (Inst), Cerulean Pharma (Inst), Genentech/Roche (Inst), GlaxoSmithKline (Inst), Leap Therapeutics (Inst), Ludwig Institute for Cancer Research (Inst), Merck (Inst), Millennium (Inst), Morab (Inst), Morphotek (Inst), Novartis (Inst), Pfizer (Inst), Syndax (Inst), and Tesaro (Inst). Patent and royalties were received by British Journal of Ob/Gyn and UpToDate.

Reference

Duska LR, Xiang Y, Hasegawa K, et al. Pembrolizumab with concurrent chemoradiotherapy in participants with high-risk locally advanced cervical cancer: a descriptive analysis of final survival from the phase 3, randomized, double-blind ENGOT-cx11/GOG-3047/KEYNOTE-A18 study. J Clin Oncol. 2025;43(suppl 17):LBA5504. doi:10.1200/JCO.2025.43.17_suppl.LBA5504