In the phase 3 KEYNOTE-564 trial (NCT03142334), adjuvant pembrolizumab (pembro) was superior to placebo (pbo) in participants (pts) with clear cell renal cell carcinoma (ccRCC) at increased risk of recurrence post nephrectomy. We report an analysis of pretreatment and on-treatment ctDNA (using 2 versions of a ctDNA assay with different sensitivities) with clinical outcomes in KEYNOTE-564.
Pts with ccRCC at increased risk of recurrence post nephrectomy were randomized 1:1 to pembro or pbo. ctDNA was evaluated using Natera’s Signatera RUO tissue-exome-based 16-plex and 64-plex ctDNA assays. Primary objectives of this analysis were to evaluate the association of baseline ctDNA status and ctDNA status change from baseline to cycle 5 day 1 (C5D1) with disease-free survival (DFS). Significance was prespecified at multiplicity-adjusted α = 0.05. Sensitivity (percentage of ctDNA-positive pts among all pts with a DFS event) and specificity (percentage of ctDNA-negative pts among all pts without a DFS event) to predict DFS were also evaluated.
994 pts were randomized to pembro (n = 496) or pbo (n = 498). ctDNA samples from 736 pts (pembro n = 374; pbo n = 362) were analyzed at baseline and from 641 pts at C5D1 (pembro n = 314; pbo n = 327) using both assays. Median study follow-up was 69.5 mo (range 60.2-86.9). In both treatment arms, ctDNA was detectable at baseline in 40 pts (5.4%) using the 16-plex assay and 60 pts (8.2%) using the 64-plex assay. By risk group, using the 16-plex assay, ctDNA was detectable at baseline in 26 of 642 (4.0%) intermediate-high risk pts, 10 of 53 (18.9%) high-risk pts, and 4 of 38 (10.5%) M1 NED pts; using the 64-plex assay, ctDNA was detectable in 44 of 642 (6.9%) intermediate-high risk pts, 11 of 53 (20.8%) high-risk pts, and 5 of 38 (13.2%) M1 NED pts. For both assays, ctDNA positivity was negatively associated with DFS in both pembro and pbo arms (P’s < 0.05). ctDNA status at baseline showed low sensitivity and high specificity to predict DFS in both treatment arms (Table). Of pts with detectable ctDNA at baseline and evaluable ctDNA at C5D1, ctDNA clearance at C5D1 occurred in 6 of 10 pts in the pembro arm and 3 of 14 pts in the pbo arm using the 16-plex assay; and 10 of 18 pts in the pembro arm and 9 of 25 pts in the pbo arm using the 64-plex assay. In both assays, ctDNA change from baseline at C5D1 was associated with DFS for both pembro and pbo arms (P's < 0.05).
While sensitivity was low, ctDNA positivity was associated with worse DFS outcomes irrespective of the ctDNA assay used. ctDNA clearance was higher in the pembro arm than the pbo arm. These data highlight the limitations of these ctDNA assays in ccRCC.