OBJECTIVES: Long-term prevention of metastatic disease remains a challenge in locally advanced rectal cancer, and robust pretreatment prognostic factors for metastatic progression are lacking. We hypothesized that detecting circulating tumor-specific DNA (ctDNA) based on hypermethylation of the neuropeptide Y gene (meth-ctDNA) could be a prognostic marker in the neoadjuvant setting; we examined this in a secondary, explorative analysis of a prospective trial. MATERIALS AND METHODS: Serum samples were prospectively collected in a phase III trial for locally advanced rectal cancer. Positivity for and fractional abundance of meth-ctDNA in baseline samples were estimated. Overall survival (OS) and the rate of distant metastases were compared between meth-ctDNA positive and negative patients; other prognostic factors were controlled for in multivariate Cox regression. Importance of quantitative load was examined by considering the fractional abundance of meth-ctDNA relative to total circulating DNA. RESULTS: Baseline serum samples were available for 146 patients. In total, 30 patients had presence of meth-ctDNA, with no correlation with cT (P=0.8) or cN (P=0.6) stages. Median follow-up was 10.6 years for OS and 5.1 years for freedom from distant metastases. Patients with meth-ctDNA had significantly worse 5-year OS (47% vs. 69%), even when controlling for other prognostic factors (hazard ratio=2.08; 95% confidence interval, 1.23-1.51). This seemed mainly driven by disparity in the rate of distant metastases (55% vs. 72% at 5 y, P=0.01); hazard ratio=2.20 (95% confidence interval, 1.19-4.07, P=0.01) in multivariate analysis. Increased quantitative load was highly significant for worse outcomes. CONCLUSIONS: Meth-ctDNA could be a potential prognostic marker in the neoadjuvant setting and may, if validated, identify patients at increased risk of distant metastases.