Background: Gastric and esophageal cancers (GECs) together account for a significant global burden in terms of new diagnoses and deaths. Over the last several years, the utility of ctDNA has ranged from estimating tumor burden and characterizing the genomic landscape of tumor biology and response to therapy in the metastatic setting to detection of minimal residual disease (MRD) and cancer surveillance in the locally advanced setting. We have previously studied the role of a commercial ctDNA assay for MRD detection and surveillance in locally advanced GECs. Herein, we present our experience with the same ctDNA assay in advanced stage settings for GECs. Methods: In this retrospective analysis of real-world data obtained from commercial circulating tumor DNA (ctDNA) testing, 53 patients with recurrent/ metastatic esophageal cancer were analyzed. A total of 216 plasma samples were collected between 10/29/2008 and 08/23/2022. The patients were divided into 3 cohorts: Cohort A (N=30) included patients with stage II/III disease who had confirmed clinical recurrence (R). Cohort B (N=25) included recurrent and Stage IV patients who achieved a state of no evidence of disease (NED) on imaging. Cohort C (N=5) included recurrent and Stage IV patients who transiently achieved NED on imaging. A personalized, tumor-informed multiplex PCR-based next-generation sequencing assay (Signatera) was used to quantify ctDNA either postoperatively, on adjuvant or palliative therapy, or during active surveillance. Results: Of 53 patients, 30 patients with recurrent stage II/III esophageal cancer had ctDNA status available postoperatively within 150 days of confirmed clinical recurrence. Cohort A: Among these cases, 25 patients were ctDNA-positive ahead of clinical recurrence (sensitivity 83.3%) with a median of 31 days (range: 1-147 days). Cohort B: Next, we explored the correlation between ctDNA status and imaging, wherein, 96% (24/25) of patients showed a significant correlation between ctDNA status (positive or negative) and disease status by imaging (Fisher exact test p=0.0001). Of the 23 patients, 17 patients were ctDNA-negative and showed NED on imaging (negative predictive value of 100%; 17/17). Cohort C: 100% of patients (N=6) who demonstrated recurrent disease on imaging after NED were ctDNA-positive prior to imaging (positive predictive value of 100%). Conclusions: Our data demonstrate the utility of ctDNA in accurately predicting disease progression and support the potential use of ctDNA to inform treatment decisions or prompt early radiographic imaging. ctDNA may ultimately supersede traditional radiographic surveillance with the advantage of being minimally invasive and cost-effective in monitoring patients during/post-treatment.