Impact of postoperative integrated genomic and epigenomic signatures of circulating tumor DNA (ctDNA) on recurrence in resected colorectal cancer: Initial report of a prospective ctDNA monitoring study COSMOS-CRC-01

168 Background: Identifying molecular residual disease (MRD) using circulating tumor DNA (ctDNA) analysis after curative surgery can potentially stratify the recurrence risk and facilitate personalization of adjuvant treatment in patients with colorectal cancer (CRC). We conducted a prospective study “COSMOS-CRC-01” to evaluate the utility of a plasma-only ctDNA assay integrating genomic and epigenomic signatures. Methods: Patients with resectable clinical stage 0–III colorectal cancer were eligible. Plasma samples were collected at structured pre- and post-surgical timepoints and analyzed using Guardant Reveal, a plasma-only ctDNA assay that detects the presence of MRD by identifying somatic alterations and methylation signatures of cell-free DNA associated with colorectal cancer. Results: As of April 2021, 501 patients were enrolled in the COSMOS-CRC-01, of which 496 patients had their post-operative 4-week ctDNA status. In this analysis, we included the first 100 patients enrolled with clinical stage II or more CRC. Seven patients were excluded due to non-curative resection or pathological stage (pStage) IV. The assay was able to produce a result for all 93 samples analyzed (failure rate of 0%). MRD was detected in 23 (25%) patients 4 weeks after surgery. The MRD detection rate was 20% in pStage II disease and 29% in pStage III disease. Patients with positive MRD were older than those with negative MRD ( p = 0.0001). Across all MRD positive samples, 30%, 9%, and 61% were positive by both genomic and epigenomic, only genomic, and only epigenomic calls. Genomic signatures included APC, BRAF, KRAS, and TP53 mutations with the lowest variant allelic fraction of 0.04%. With a median follow-up time of 12.2 months (range 8-18 months), 9 of 93 patients recurred (9.7%). ctDNA was detected from a single 4-week post-surgical sample in 55% of patients who recurred (5 of 9). 1-year disease-free survival was 81.2% in patients with positive MRD and 93.9% in those with negative MRD (hazard ratio 3.49, 95% CI 0.93–13.10, p = 0.049). Multivariate analysis including baseline characteristics associated with recurrence risk showed that MRD status had the strongest association with the recurrence. Post-operative 4-week serum CEA level was not associated with risk for recurrence. Conclusions: While follow-up in this cohort is currently limited, the results suggest that a single post-surgery sample run on a plasma-only assay that integrates genomic and epigenomic signatures can more accurately stratify patients with CRC by recurrence risk than previously known clinical factors. Data from baseline and longitudinal timepoints will be reported as available along with longer term clinical follow-up. Clinical trial information: UMIN000037765.