In brief
- At five years, event-free survival in KEYNOTE-671 was 49.9% (95% CI 44.6 to 55.0) with perioperative pembrolizumab and 26.5% (21.7 to 31.5) with placebo, a hazard ratio of 0.58 (0.48 to 0.69).
- Five-year overall survival was 64.6% (59.5 to 69.2) against 53.6% (48.3 to 58.6), a hazard ratio of 0.74 (0.59 to 0.92).
- 797 patients with previously untreated, resectable stage II, IIIA or IIIB (N2) NSCLC were randomised, 397 to pembrolizumab and 400 to placebo.
- Median time from randomisation to the 3 July 2025 data cutoff was 60.4 months, with a range of 42.6 to 85.8 months.
- No detriment to health-related quality of life was identified with the longer follow-up, and safety was consistent with the known profiles of each treatment.
Study at a glance
- Cohort: KEYNOTE-671 (ClinicalTrials.gov NCT03425643), phase 3, randomised, 797 participants
- Population: Previously untreated, resectable stage II, IIIA or IIIB (N2) non-small cell lung cancer
- Treatment: Four cycles of pembrolizumab 200 mg or placebo every three weeks with platinum-doublet chemotherapy, then surgery, then pembrolizumab or placebo every three weeks for up to 13 cycles
- Primary endpoints: Event-free survival by RECIST v1.1 investigator assessment, and overall survival
- Follow-up: Median 60.4 months from randomisation to the 3 July 2025 cutoff (range 42.6 to 85.8)
- Five-year event-free survival: 49.9% against 26.5% (hazard ratio 0.58, 95% CI 0.48 to 0.69)
- Five-year overall survival: 64.6% against 53.6% (hazard ratio 0.74, 95% CI 0.59 to 0.92)
- Not reported in the abstract: Median event-free or overall survival in either arm, and any subgroup result
Five-year event-free survival of 49.9% against 26.5%
Half of the patients randomised to perioperative pembrolizumab were alive without an event at five years. In the placebo group the figure was 26.5%. The hazard ratio of 0.58, with a 95% confidence interval from 0.48 to 0.69, does not cross 1, and the confidence intervals around the two five-year rates (44.6 to 55.0 and 21.7 to 31.5) do not overlap.
Event-free survival here was assessed by the investigator using RECIST v1.1, which the authors state in the methods, and it is one of two primary endpoints rather than a secondary measure. Overall survival is the other, so both of the trial’s primary questions are answered in this report at the same cutoff.
An 11 percentage point difference in five-year overall survival
Overall survival at five years was 64.6% with pembrolizumab and 53.6% with placebo, a hazard ratio of 0.74 with a 95% confidence interval from 0.59 to 0.92. The interval excludes 1, and the separation the authors describe as continuing represents 11 percentage points at the five-year mark.
The overall survival difference is smaller in proportional terms than the event-free survival difference. Event-free survival ran at 49.9% against 26.5%, a ratio of about 1.9, while overall survival ran at 64.6% against 53.6%, a ratio of about 1.2. The abstract records no data on what patients received after recurrence, so what lies behind that narrowing cannot be read from it.
“After 5 years of follow-up, EFS was nearly double with perioperative pembrolizumab plus neoadjuvant chemotherapy, along with continued improvement in OS compared with neoadjuvant chemotherapy and surgery alone.”Wakelee et al., Annals of Oncology (2026)
A median 60.4 months of follow-up, in a range running from 42.6 to 85.8
Median time from randomisation to the data cutoff of 3 July 2025 was 60.4 months, and the range ran from 42.6 to 85.8 months. The lower end of that range matters to the reading of a five-year rate: the patients randomised latest had been followed for 42.6 months at the cutoff, which is short of five years, so not every participant contributed observed follow-up to the five-year timepoint.
Patients received four cycles of pembrolizumab 200 mg or placebo every three weeks with platinum-doublet chemotherapy, then surgery, then up to 13 cycles of pembrolizumab or placebo every three weeks. Randomisation was 1:1, and 797 patients were assigned, 397 to pembrolizumab and 400 to placebo.
No quality-of-life detriment, and safety consistent with the known profiles
The authors report that no detriment to health-related quality of life was identified in the pembrolizumab arm against the placebo arm with the longer follow-up, and that safety was consistent with the known profiles of each treatment. The abstract carries no adverse event rates, no instrument name for the quality-of-life assessment and no timepoints for it.
On these results the authors state that the durable and clinically meaningful benefits observed support use of this treatment as a standard of care for patients with resectable early-stage non-small cell lung cancer. That sentence is the authors’ own conclusion rather than a finding, and it rests on the two five-year rates and the quality-of-life and safety statements above.
Sources
- Annals of Oncology. Five-Year Outcomes of Perioperative Pembrolizumab for Early-Stage Non–Small-Cell Lung Cancer From the Randomized KEYNOTE-671 Study (2026-08-01). doi.org
Featuring Lung Summit faculty
This study was co-authored by Lung Summit faculty Delvys Rodríguez-Abreu and Martin Reck.
This article was produced independently by the Lung Summit editorial team, without industry funding or input.
