Clinical utility of circulating tumor DNA (ctDNA) for participants with resected, high-risk dMMR endometrial cancer (EC) receiving adjuvant treatment: Results from the ENGOT-en11/GOG-3053/KEYNOTE-B21 study

KEYNOTE-B21

Patient / Population Intervention / Exposure Comparison Outcome

In participants with resected, high-risk dMMR endometrial cancer, adjuvant pembrolizumab plus chemotherapy favoured disease-free survival compared with placebo plus chemotherapy, while ctDNA status had only modest predictive value for disease-specific-free survival.

N

—patients

Design

Randomised 1:1, blinded, placebo-controlled trial of adjuvant pembrolizumab or placebo plus standard-of-care treatment in participants with high-risk endometrial cancer after curative intent surgery, reported here for the dMMR subgroup with prospective ctDNA assessment by the Signatera 16-plex assay (NCT04634877).

Endpoint

Disease-free survival (primary endpoint). Disease-specific-free survival and its association with ctDNA status were exploratory.

ResultIn dMMR endometrial cancer at IA4, DFS HR 0.34 (95% CI 0.17-0.65) with pembrolizumab plus chemotherapy vs placebo plus chemotherapy, below the prespecified HR boundary of 0.4161. DSFS HR 0.25 (95% CI 0.11-0.54). In the ITT population at IA3, pembrolizumab did not improve DFS or DSFS. Baseline ctDNA positivity was 21% vs 19% (n = 182). At Stage 2 C2D1, ctDNA was maintained positive in 1% vs 6% and cleared in 20% vs 14%. At week 96, 0% vs 5% were maintained and 17% vs 10% cleared. Baseline ctDNA predicted DSFS with sensitivity 50% vs 21%, specificity 80% vs 82%, PPV 11% vs 24% and NPV 97% vs 80%.

Backes F, Ortaç F, Kim YM, et al. Presented at ESMO Congress 2026, abstract LBA36, 24 October. Not yet published.

Discussion & critique

In the dMMR subgroup, adjuvant pembrolizumab was associated with fewer DFS and DSFS events at a fourth interim analysis, and the DFS HR crossed a prespecified boundary. The overall trial did not improve DFS in the ITT population, so this is a subgroup finding. The abstract gives no number randomised in the dMMR subgroup, no event counts, medians, P values or safety data. The ctDNA analysis was exploratory and limited to participants with samples. Post-baseline numbers fell to 82 by week 96 and included only participants who completed planned therapy. With low ctDNA positivity and few DSFS events, the assay's low PPV and variable sensitivity limit its use to guide adjuvant decisions, and the authors themselves call for more sensitive methods.