Exome-based ctDNA shows low sensitivity for recurrence detection in adjuvant ccRCC

Exome-based circulating tumor DNA (ctDNA) positivity at baseline was associated with worse disease-free survival (DFS) to both adjuvant pembrolizumab (Keytruda) and placebo among patients with high-risk clear cell renal cell carcinoma (ccRCC), although the assays demonstrated low sensitivity for recurrence detection, according to an analysis of the phase 3 KEYNOTE-564 trial (NCT03142334).¹

The data were presented by Toni K. Choueiri, MD, FASCO, at the 2026 American Society of Clinical Oncology Annual Meeting in Chicago, Illinois.

“In my opinion, these data highlight the limitation of current testing with exome-based ctDNA, not ctDNA in general, in adjuvant clear cell RCC,” Choueiri noted during the presentation.

Background on KEYNOTE-564

KEYNOTE-564 previously demonstrated that adjuvant pembrolizumab significantly improved DFS (HR, 0.71) and overall survival (OS; HR, 0.66) vs placebo in patients with high-risk RCC following nephrectomy. The present analysis evaluated whether ctDNA could identify patients at highest risk for recurrence or predict benefit from adjuvant therapy.

In the double-blind phase 3 trial, 994 patients with ccRCC were randomly assigned 1:1 to pembrolizumab (n = 496) or placebo (n = 498) for up to 1 year (or 17 cycles). ctDNA analyses were conducted using a modified version of the Natera Signatera RUO tissue-exome-based cDNA assay in patients who had tumor tissue, whole blood, and plasma samples available. Investigators evaluated both a 16-plex assay approximating commercially available testing and a more sensitive 64-plex exome-based platform.

Findings from the study

Baseline ctDNA samples from 736 patients were evaluable, and 641 patients had paired samples available at cycle 5 day 1. The median follow-up was 69.5 months (range, 60.2 to 86.8).

Across both treatment groups, detectable ctDNA at baseline correlated with inferior DFS (both P < .05) irrespective of assay used.

However, the overall detection rate was low. Using the 16-plex assay, ctDNA was detectable in 4.0% of patients with intermediate-high–risk disease, 18.9% of patients with high-risk disease, and 10.5% of patients with M1 no evidence of disease. Corresponding rates with the 64-plex assay were 6.9%, 20.8%, and 13.2%, respectively.

Although specificity exceeded 95% across analyses, sensitivity for recurrence prediction remained poor. Baseline ctDNA sensitivity ranged from 10% to 15% depending on assay and treatment arm. Positive predictive values ranged from 73% to 95%. The negative predictive values were 55% (16-plex) and 56% (64-plex) in the pembrolizumab arm, and 63% (16-plex) and 64% (64-plex) in the placebo arm.

Investigators also evaluated ctDNA kinetics during therapy. Among patients with detectable ctDNA at baseline who remained evaluable during treatment, ctDNA clearance occurred more frequently with pembrolizumab than placebo. Using the 16-plex assay, clearance occurred in 6 of 10 (60%) evaluable patients in the pembrolizumab arm compared with 3 of 14 (21.4%) patients in the placebo arm. With the 64-plex assay, clearance occurred in 10 of 18 (55.6%) vs 9 of 25 (36%) patients, respectively. With both assays, ctDNA change from baseline to cycle 5 day 1 was associated with DFS for both pembrolizumab and placebo (both P < .05)

According to Choueiri, these results do not support the routine use of currently available ctDNA technology (exome-based) to guide adjuvant strategies in this setting.

He concluded during the presentation, “These data highlight the limitations of [low] ctDNA positivity rate in these high-risk RCC patients [who] are currently using something that I would say, in theory, will not be used. This is an example of the science getting way ahead, but we're still doing things in the past.”

REFERENCE

1. Choueiri TK. ctDNA analysis in participants with renal cell carcinoma treated with adjuvant pembrolizumab or placebo in the KEYNOTE-564 trial. J Clin Oncol. 44, 2026 (suppl 16; abstr 4502)