Short answer: KEYNOTE-826 was a double-blind phase 3 trial of 617 women with persistent, recurrent or metastatic cervical cancer not amenable to curative treatment. Adding pembrolizumab to platinum and paclitaxel, with or without bevacizumab, improved progression-free survival (median 10.4 vs 8.2 months in the PD-L1 CPS 1 or higher population) and overall survival. In the final analysis, median OS in the whole population was 26.4 vs 16.8 months (HR 0.63), and 28.6 vs 16.5 months in patients with CPS 1 or higher (HR 0.60). The benefit came with more immune-related adverse events (33.9% vs 15.2%). This guide covers the design, the numbers and the subgroup caveats that matter for boards and for clinic.
Author and disclosure: Roupen Odabashian, MD, FRCPC, Oncologist and founder of MeDucation AI. I am the founder of MeDucation AI, which sells hematology and oncology board preparation. I have no other financial relationships relevant to the studies discussed here. This article is educational and is not a substitute for individualized clinical judgment or tumor board discussion.
What was the KEYNOTE-826 trial design?
KEYNOTE-826 was a randomized, double-blind, placebo-controlled phase 3 trial conducted at 151 sites in 19 countries and funded by Merck Sharp and Dohme. Patients had persistent, recurrent or metastatic cervical cancer that was not amenable to curative treatment, had received no prior systemic chemotherapy for advanced disease, and had an ECOG performance status of 0 or 1. They received paclitaxel plus cisplatin or carboplatin, with or without bevacizumab at the investigator's discretion (about 63% in each arm), plus either pembrolizumab or placebo. Randomization was stratified by metastatic status at diagnosis, planned bevacizumab use and PD-L1 combined positive score (CPS). The dual primary endpoints were progression-free survival (PFS) and overall survival (OS) in three nested populations: CPS 1 or higher (about 88% of patients), the intention-to-treat population, and CPS 10 or higher (about 51%). The primary report was by Colombo and colleagues in the New England Journal of Medicine in 2021.
What were the efficacy results?
Population | Median PFS (pembrolizumab vs placebo) | Objective response rate |
|---|---|---|
CPS 1 or higher | 10.4 vs 8.2 months | 68.1% vs 50.2% |
All patients (ITT) | 10.4 vs 8.2 months | 65.9% vs 50.8% |
CPS 10 or higher | 10.4 vs 8.1 months | 69.6% vs 49.1% |
All PFS and OS comparisons in the primary analysis were statistically significant (P < 0.001) at a median follow-up of 22 months. Overall, pembrolizumab reduced the risk of progression or death by 35% and the risk of death by 33%, and the duration of response was roughly twice as long.
What did the final overall survival analysis show?
Population | Median OS (pembrolizumab vs placebo) | Hazard ratio (95% CI) |
|---|---|---|
All patients (n = 617) | 26.4 vs 16.8 months | 0.63 (0.52 to 0.77) |
CPS 1 or higher | 28.6 vs 16.5 months | 0.60 (0.49 to 0.74) |
CPS 10 or higher | 29.6 vs 17.4 months | 0.58 (0.44 to 0.78) |
The final analysis, reported by Monk and colleagues in the Journal of Clinical Oncology, had a median follow-up of 39.1 months at the October 2022 data cutoff and confirmed the survival benefit seen at the interim analysis.
Does the benefit hold in subgroups and without bevacizumab?
An exploratory subgroup analysis found OS hazard ratios favoring pembrolizumab in every subgroup, but with wide confidence intervals in some. Treat these as descriptive rather than definitive.
Subgroup (CPS 1 or higher population) | OS hazard ratio (95% CI) |
|---|---|
With bevacizumab | 0.62 (0.45 to 0.87) |
Without bevacizumab | 0.67 (0.47 to 0.96) |
Squamous histology | 0.60 (0.46 to 0.79) |
Non-squamous histology | 0.70 (0.41 to 1.20) |
Prior chemoradiotherapy only | 0.56 (0.39 to 0.81) |
No prior chemoradiotherapy | 0.72 (0.52 to 1.00) |
Carboplatin | 0.65 (0.50 to 0.85) |
Cisplatin | 0.53 (0.27 to 1.04) |
Several intervals cross 1.0, including cisplatin and non-squamous histology, so those estimates are imprecise. The authors note that non-squamous tumors are more likely to be PD-L1 negative, and that bevacizumab should be considered when available unless contraindicated, while pembrolizumab plus chemotherapy still offers meaningful benefit without it.
What are the safety signals?
Grade 3 or higher adverse events: 81.8% with pembrolizumab vs 75.1% with placebo.
Serious adverse events: 49.8% vs 42.4%.
Most common grade 3 or higher events: anemia (30.6% vs 26.9%) and neutropenia (12.4% vs 9.7%).
Immune-related adverse events: 33.9% vs 15.2%, mostly grade 1 or 2 thyroid disorders.
Board-style takeaway: the adverse event profile mixes chemotherapy and bevacizumab toxicity with immune-related toxicity, so hypothyroidism and other endocrinopathies should be monitored during and after treatment.
What should you remember for the boards?
KEYNOTE-826 is the first-line trial supporting pembrolizumab plus chemotherapy, with or without bevacizumab, in persistent, recurrent or metastatic cervical cancer.
Know the PD-L1 CPS populations (1 or higher, 10 or higher, ITT) and that the final OS hazard ratios were roughly 0.6 across them.
The authors favor adding bevacizumab where feasible, but the pembrolizumab benefit persisted without it.
Check the current prescribing information for the approved PD-L1 population rather than relying on the trial's ITT results.
To practice this style of question, see Oncology Board Review Question Banks Compared, or build a plan with How to Pass the ABIM Medical Oncology Boards. For the wider exam picture, read Oncology Board Review 2026.
Frequently asked questions
What did KEYNOTE-826 show?
Adding pembrolizumab to chemotherapy, with or without bevacizumab, improved progression-free survival and overall survival in first-line persistent, recurrent or metastatic cervical cancer, with a final OS hazard ratio of 0.63 in all patients.
Does PD-L1 status matter in KEYNOTE-826?
The trial tested three populations: CPS 1 or higher, all patients and CPS 10 or higher. Benefit was seen in all, with final OS hazard ratios of 0.60, 0.63 and 0.58, and the approved population should be confirmed in the current label.
Is bevacizumab required with pembrolizumab in cervical cancer?
No. About 63% of patients received bevacizumab, and the survival benefit of pembrolizumab appeared in subgroups with and without it, although the authors recommend considering bevacizumab when available and not contraindicated.
What are the main side effects of adding pembrolizumab?
Immune-related adverse events occurred in 33.9% vs 15.2%, mostly low-grade thyroid disorders, on top of the anemia and neutropenia expected with chemotherapy.
Which cervical cancer histologies were studied?
Squamous and non-squamous tumors were both included. Squamous disease showed a clearer survival hazard ratio, while the non-squamous estimate was imprecise.
References
Colombo N, Dubot C, Lorusso D, et al. Pembrolizumab for persistent, recurrent, or metastatic cervical cancer. N Engl J Med. 2021. ESMO Congress 2021 abstract record
Goodman A. Study shows pembrolizumab plus chemotherapy improves survival in advanced cervical cancer. The ASCO Post. October 10, 2021. ascopost.com
Monk BJ, et al. Addition of first-line pembrolizumab to chemotherapy with or without bevacizumab in advanced cervical cancer: final overall survival results from KEYNOTE-826. J Clin Oncol. 2023. doi:10.1200/JCO.23.00914
The ASCO Post. Final overall survival results from KEYNOTE-826. November 2023. ascopost.com
ESMO. Adding pembrolizumab to chemotherapy with or without bevacizumab improves overall survival across subgroups of patients with persistent, recurrent or metastatic cervical cancer. esmo.org
Tewari KS, Colombo N, Monk BJ, et al. Subgroup analyses from KEYNOTE-826. J Clin Oncol. 2022;40(suppl 16):5506. Summary: CancerNetwork
ClinicalTrials.gov. KEYNOTE-826, NCT03635567. clinicaltrials.gov
Medicom. Pembrolizumab prolongs survival in persistent, recurrent or metastatic cervical cancer (ESMO 2021). conferences.medicom-publishers.com
