Background: Markers that better define recurrence risk for patients (pts) with stage II CC are urgently required, potentially defining a subset that would most benefit from adjuvant chemotherapy (CTX) and intensive surveillance. An alternative strategy to standard analyses of the resected surgical specimen is to directly examine plasma for evidence of residual disease. Recently, ctDNA has shown promise as a blood biomarker in advanced colorectal cancer; here we explore the potential of this marker in stage II CC. Methods: In this prospective study, plasma samples are being collected at 4-10 weeks post-op in 250 stage II CC pts, with serial 3 monthly samples on a subset of 175 pts. Adjuvant CTX is at clinician discretion, blinded to ctDNA analysis. Surveillance includes 3 monthly CEA and 6 monthly CT imaging for 2 years. All samples were sent to Johns Hopkins Kimmel Cancer Center, where tumor tissue was analyzed for hotspot mutations in TP53, APC, KRAS, NRAS, BRAF, PIK3CA, CTNNB1, SMAD4, and FBXW7 using a massively parallel sequencing platform (Safe-SeqS). The identified mutation was queried and quantified in plasma using the same platform, blinded to clinical data. Results: Preliminary data is available on 78 of the 190 pts enrolled to date. Median age is 66 years, 25/78 (32%) received adjuvant CTX. At least 1 mutation was found in all tumors, with matching ctDNA detectable in 6/78 (7.7%) plasma samples. 10 (12.8%) recurrences have occurred at a median follow-up of 507 days, including 5 of 6 pts with detectable ctDNA and 5 of 72 with no detectable ctDNA (Table). Pts with detectable ctDNA had a shorter recurrence-free survival (median 234 days vs undefined, HR 23.09, log-rank p < 0.0001). In an exploratory analysis of the correlation between ctDNA and clinicopathologic features, detectable ctDNA maintained prognostic significance, including for T3 tumors (HR for time to recurrence 80.55, log-rank p < 0.0001). Conclusions: Early data suggests ctDNA is a promising marker of recurrence risk in stage II CC. Preliminary analyses suggests this may be independent of clinicopathologic features.

Fisher’s exact P < 0.0001. RR = 12 (95% CI 4.79 – 30.06).