BERLIN — The combination of pembrolizumab with weekly paclitaxel, with or without bevacizumab, leads to significant improvements in survival outcomes in patients with platinum-resistant recurrent ovarian cancer (PRROC), demonstrates KEYNOTE-B96.

The research, presented by Nicoletta Colombo, MD, at the European Society for Medical Oncology (ESMO) Congress 2025 on October 18, showed that the regimen was associated with progression-free survival (PFS) and overall survival improvements in patients with a programmed death ligand 1 (PD-L1) combined positive score (CPS) ≥ 1. The study funder, Merck, recently announced in a press release that pembrolizumab was also associated with a significant improvement in overall survival in the intention-to-treat (all-comers) population.

“This is the first phase 3 study to report statistically significant improvement in overall survival with an immune checkpoint inhibitor-based regimen in ovarian cancer,” said Colombo, of the University of Milan–Bicocca, European Institute of Oncology Istituto di Ricovero e Cura a Carattere Scientifico, Milan. She also described the benefit as “clinically meaningful.”

Experts Divided on Whether Trial Suggests New SOC

With no new safety signals, Colombo said, “These data support the use of pembrolizumab plus weekly paclitaxel, with or without bevacizumab, as a new standard of care for patients with PRROC.”

Study discussant Isabelle L. Ray-Coquard, MD, PhD, oncologist at Centre Léon Bérard, and professor of medicine at Université Claude Bernard Lyon I, Lyon, France, who was not involved in the study, said that not all patients are able to have bevacizumab because of either prior exposure or bowel infiltration.

However, for those patients who are candidates for weekly paclitaxel plus or minus bevacizumab, the addition of pembrolizumab should be proposed for those with PD-L1 CPS ≥ 1, she continued.

“Together with advances such as mirvetuximab [soravtansine] and relacorilant, this study expands treatment choices for women with relapsed ovarian cancer, a group with historically poor outcomes.”

Ray-Coquard said that “future priorities” for research should include biomarker-driven patient selection, optimal treatment sequencing, and ensuring equitable access to testing and innovation for all women.

“But today we have to remember we have a new strategy available for our patients.”

On X, Kathleen N. Moore, MD, MS, deputy director and co-director of the Cancer Therapeutics Program at the Stephenson Cancer Center, University of Oklahoma, Norman, noted the improved survival outcomes but wondered how the results compare with those for mirvetuximab soravtansine and nab-paclitaxel/relacorilant.

“And with the tsunami of [antibody–drug conjugates], does this change the SOC?” she asked. Her answer: “Not sure.”

Taro Yamanaka, MD, a medical oncologist at Japan's National Cancer Center Hospital, added on X that there are “still many questions” to be answered, such as over the identification of further biomarkers, the use of bevacizumab, prior poly(ADP)-ribose polymerase (PARP) inhibitor therapy, and treatment sequencing.

“But great to see new options emerging!”

Methods and Results

Colombo began her presentation by noting that the phase 3 AURELIA trial showed that a combination of bevacizumab and weekly paclitaxel had the greatest efficacy among the chemotherapy regimens included in the study.

She explained that paclitaxel boosts immunity by inducing cell death, reshaping the tumor microenvironment, enhancing antigen presentation, and reducing regulatory T cells, and that combining it with pembrolizumab enhances the anti-tumor immune response and thus could further improve outcomes.

The researchers undertook KEYNOTE-B96 in patients with histologically confirmed epithelial ovarian, fallopian tube, or primary peritoneal carcinoma.

The patients were required to have received one or two prior lines of therapy, at least one of which had to be platinum-based chemotherapy. Prior immune checkpoint and PARP inhibitors and bevacizumab were permitted. They also had to have radiographic progression within 6 months of the last dose of platinum-based chemotherapy.

They were then randomly assigned to pembrolizumab or placebo for 18 cycles, alongside paclitaxel on days 1, 8, and 15 of each 3-week cycle, with or without bevacizumab every 2 weeks.

Colombo reported that, of 867 patients screened, 643 from 187 sites in 25 countries were randomly assigned: 322 to pembrolizumab plus paclitaxel and 321 to placebo plus paclitaxel.

The median age of the patients was 62 years in the pembrolizumab arm and 61 in the group given placebo. Approximately 63% had received two prior lines of therapy, with PARP inhibitors used in around 36%, and just over 45% had previously received bevacizumab. As part of the study, bevacizumab was given to approximately 73% of patients.

Colombo also highlighted that there was a range of PD-L1 expression, with around 27% having CPS < 1, just over 41% a score of 1 to < 10, and 31% a score of ≥ 10.

For the primary endpoint of PFS, the median follow-up was 15.6 months. Colombo reported that, among participants with a CPS score ≥ 1, median PFS was 8.3 months in the pembrolizumab arm vs 7.2 months in patients given placebo, at a hazard ratio of 0.72 (P = .0014). 

“This difference is very similar to what we observed in recent trials in the same population,” she said, although she noted that, here, the control arm performed better than in those earlier studies.

In the overall intention-to-treat, all-comer population, the median PFS was again 8.3 months with pembrolizumab vs 6.4 months in the placebo arm, at a hazard ratio of 0.70 (P < .0001).

“The benefit observed in the overall population was confirmed across different prespecified subgroups,” Colombo noted, “and this was true for the CPS 1 and higher population, and also the intention-to-treat population,” and “regardless of the use of bevacizumab.”

Turning to the key secondary endpoint of overall survival, she explained that the median follow-up was 26.6 months. In the CPS ≥ 1 population, the median overall survival was 18.2 months with pembrolizumab vs 14.0 months with placebo, at a hazard ratio of 0.76 (P = .0053). Again, the benefit was seen across key subgroups, and regardless of prior therapy.

She then showed that pembrolizumab was associated with a higher objective response rate, at 50.4% vs 40.8% in the placebo arm, or a delta of 8.9%, and had more complete responses, at 8.3% vs 6.0%. The duration of response was also longer in the experimental arm, with 26.5% still responding at 18 months vs 14.5% in patients assigned to placebo.

More Patients in Pembrolizumab Arm Had TRAEs, Serious AEs

As expected, the proportion of patients experiencing treatment-related adverse events (TRAEs) was higher in the pembrolizumab arm, with 67.5% experiencing a grade ≥ 3 event, vs 55.3% with placebo, and 11.6% vs 3.5% having a grade ≥ 3 immune-mediated adverse event.

Serious adverse events were also more common with pembrolizumab, as were adverse events leading to discontinuation of any treatment. However, the rate of adverse events leading to death was similar in the two arms.

The most common TRAEs were anemia (49.7% with pembrolizumab vs 42.1% on placebo), peripheral neuropathy (38.8% vs 31.1%), and alopecia (37.8% vs 34.0%).

Ray-Coquard said that this trial succeeded because it was large and international, and because of the choice of weekly paclitaxel as the backbone. Also, she pointed to the addition of bevacizumab and the use of biomarker enrichment through PD-L1 CPS ≥ 1.

She noted, however, that this is an increasingly crowded space, with not only mirvetuximab soravtansine (via the MIRASOL trial) and nab-paclitaxel plus relacorilant (ROSELLA), but also several antibody-drug conjugates in late development, but that differences in the study populations makes cross-trial comparisons impossible.

Ray-Coquard also highlighted that different biomarkers were selected between KEYNOTE-B96, MIRASOL, and ROSELLA, which begs the question as to the overlap between them in terms of expression.

“In this setting,” she continued, "we need many more options,” while understanding that the sequencing of the drugs “will require new methodologies and, more importantly, collaborative, prospective studies.”

The study was funded by Merck Sharp & Dohme LLC.

Colombo declares relationships with AstraZeneca, Eisai, MSD, Merck & Co., Inc., Roche, GlaxoSmithKline, Immunogen, Eisai, OncXerna, Nuvation Bio, Gilead, Regeneron, Novocure, Seagen, AbbVie, Lily, and BeOne.